tag:old.nabble.com,2006:forum-3701Nabble - Aqsis Renderer2009-11-29T01:28:27ZAqsis is a Renderman(tm) compliant 3D rendering toolkit. Aqsis is based on the Reyes rendering architecture. Features include : Programmable Shading True Displacements NURBS CSG Motion Blur Subdivision Surfaces Aqsis Renderer home is <a href="http://sourceforge.net/projects/aqsis/" target="_top" rel="nofollow">here</a>.tag:old.nabble.com,2006:post-26560400Re: Compiling Aqsis on Snow Leopard (OS X 10.6.x)2009-11-29T01:28:27Z2009-11-29T01:28:27ZJonathan Merritt-2Hi Everyone,
<br><br>On 24/11/2009, at 8:41 AM, Matthias Baas wrote:
<div class='shrinkable-quote'><br>>> I was hoping to do something more general initially... a layered approach:
<br>>>
<br>>> Piqsl XML-based socket communication
<br>>> |
<br>>> Data layer of interfaces containing
<br>>> bucket data, listeners that are
<br>>> notified when buckets become
<br>>> available, etc.
<br>>> |
<br>>> Display layer; components in various
<br>>> toolkits (Swing, JavaFX, SWT) for
<br>>> displaying the data layer.
<br>>>
<br>>> With an appropriate "model" (in the MVC arrangement) that includes
<br>>> listeners, an adaptor-style hookup to any of the GUI toolkits should
<br>>> be easy.
</div><br>Just a quick update on this... I've implemented most of the bottom layer in a "rough draft" form. I've implemented it around a "responder" interface for socket communication that looks like this:
<br><br>public interface PiqslResponder {
<br> List<Format> openDisplay(OpenMessage message);
<br> void receiveBucket(BucketData data);
<br> void closeDisplay();
<br> void commError(String message);
<br>}
<br><br>The responder is passed to a socket manager, which performs the socket communication and asynchronously calls the various methods when required. The other items (OpenMessage, Format, BucketData, and a few others referenced by those) are also interfaces to read-only data structures.
<br><br>It's a bit rough at this stage, but everything is working. I'll probably spend the remainder of next week tidying up the socket layer. Then I'll implement a model which will be suitable for any of the GUI toolkits.
<br><br>I'm using Scala for all of this, but that should be invisible to clients of each layer. Each layer is only made visible outside of its own package via a set of interfaces and a few stub classes which are written in Java. Since Java is (currently) the lowest-common-denominator among the JVM languages, it should be possible for clients of each layer to be written in any JVM language.
<br><br>> Does that mean I could write applets or Eclipse plugins in one of the
<br>> above languages instead of Java?
<br><br>I think so, yes... but you'd need to check the specifics for how each of the languages inter-operates with Java. I know, for example, that it's possible to use Swing and SWT from Jython. Jython uses reflection (at least, last time I checked) to call appropriate methods on the Java classes.
<br><br>I can keep working on things locally, and perhaps prepare a screencast or something when I have a GUI layer finished (which wouldn't be until mid-January, probably). If anyone else wants to see what I'm up to, I can check code in to aqsis svn or perhaps to somewhere like Project Kenai (I'm using Netbeans for development, which integrates nicely with Kenai).
<br><br>There hasn't been a need to modify piqsldisplay.cpp yet. For development, I'm launching the socket listener first, and then invoking aqsis. Provided that I use:
<br> Display "name.tif" "piqsl" ...
<br>in a RIB file, aqsis will still launch piqsldisplay, and piqsldisplay will not attempt to re-launch the FLTK GUI provided that someone is already listening on the correct port. Obviously, something will need to be done to launch the GUI eventually, but tinkering with that can wait until I have a working GUI.
<br><br>Jonathan Merritt.
<br><br><br>------------------------------------------------------------------------------
<br>Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>trial. Simplify your report design, integration and deployment - and focus on
<br>what you do best, core application coding. Discover what's new with
<br>Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26560400&i=0" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-26548522Re: memset usage in d_sdcBMP.cpp2009-11-27T14:12:45Z2009-11-27T14:12:45ZChris Foster-5On Fri, Nov 27, 2009 at 10:46 PM, Paul Gregory <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26548522&i=0" target="_top" rel="nofollow">pgregory@...</a>> wrote:
<br>> Looks like a real bug, this needs to be recorded on the bug list so that it
<br>> gets fixed asap.
<br><br>Fixed now in git.
<br><br>> Only positive is that d_sdcWin32 and d_sdcBMP are pretty much unused as far
<br>> as I know.
<br><br>The bmp display driver is built and installed but I'd guess that pretty much
<br>nobody uses it. OTOH, the win32 display driver isn't even built so that bug
<br>can't be affecting anybody ;-)
<br><br>Cheers,
<br>~Chris
<br><br>------------------------------------------------------------------------------
<br>Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>trial. Simplify your report design, integration and deployment - and focus on
<br>what you do best, core application coding. Discover what's new with
<br>Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26548522&i=1" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-26543033Re: memset usage in d_sdcBMP.cpp2009-11-27T04:46:14Z2009-11-27T04:46:14ZPaul Gregory-4Looks like a real bug, this needs to be recorded on the bug list so that it gets fixed asap.<br><br>Only positive is that d_sdcWin32 and d_sdcBMP are pretty much unused as far as I know.<br><br><br>Cheers<br><br>Paul<br><br>
<div class="gmail_quote">2009/11/27 Nicolas Chauvet <span dir="ltr"><<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26543033&i=0" target="_top" rel="nofollow">kwizart@...</a>></span><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Hello,<br>
<br>
Dave Jones from the fedora kernel team has checked several userland<br>
package for incorrect usage of the memset function.<br>
Aqsis is one of them. From his report one can see:<br>
<br>
Checking ./aqsis/aqsis-1.6.0/tools/displays/sdcWin32/d_sdcWin32.cpp<br>
Found memset with swapped arguments.<br>
250: memset(&g_Data, sizeof(AppData), 0);<br>
<br>
Checking ./aqsis/aqsis-1.6.0/tools/displays/sdcBMP/d_sdcBMP.cpp<br>
Found memset with swapped arguments.<br>
172: memset(&g_Data, sizeof(AppData), 0);<br>
<br>
This seems always to be a bug when the third argument is 0.<br>
<br>
Thx<br>
<br>
Nicolas (kwizart)<br>
<br>
------------------------------------------------------------------------------<br>
Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day<br>
trial. Simplify your report design, integration and deployment - and focus on<br>
what you do best, core application coding. Discover what's new with<br>
Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_blank" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>
_______________________________________________<br>
Aqsis-development mailing list<br>
<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26543033&i=1" target="_top" rel="nofollow">Aqsis-development@...</a><br>
<a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_blank" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br>
</blockquote></div><br>
<br />------------------------------------------------------------------------------
<br>Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>trial. Simplify your report design, integration and deployment - and focus on
<br>what you do best, core application coding. Discover what's new with
<br>Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br />_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26543033&i=2" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-26541854memset usage in d_sdcBMP.cpp2009-11-27T04:34:30Z2009-11-27T04:34:30ZNicolas ChauvetHello,
<br><br>Dave Jones from the fedora kernel team has checked several userland
<br>package for incorrect usage of the memset function.
<br>Aqsis is one of them. From his report one can see:
<br><br>Checking ./aqsis/aqsis-1.6.0/tools/displays/sdcWin32/d_sdcWin32.cpp
<br>Found memset with swapped arguments.
<br>250: memset(&g_Data, sizeof(AppData), 0);
<br><br>Checking ./aqsis/aqsis-1.6.0/tools/displays/sdcBMP/d_sdcBMP.cpp
<br>Found memset with swapped arguments.
<br>172: memset(&g_Data, sizeof(AppData), 0);
<br><br>This seems always to be a bug when the third argument is 0.
<br><br>Thx
<br><br>Nicolas (kwizart)
<br><br>------------------------------------------------------------------------------
<br>Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>trial. Simplify your report design, integration and deployment - and focus on
<br>what you do best, core application coding. Discover what's new with
<br>Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26541854&i=0" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-26486445Re: Compiling Aqsis on Snow Leopard (OS X 10.6.x)2009-11-23T13:41:44Z2009-11-23T13:41:44ZMatthias Baas-3Hi Jonathan,
<br><div class='shrinkable-quote'><br>> I was hoping to do something more general initially... a layered approach:
<br>>
<br>> Piqsl XML-based socket communication
<br>> |
<br>> Data layer of interfaces containing
<br>> bucket data, listeners that are
<br>> notified when buckets become
<br>> available, etc.
<br>> |
<br>> Display layer; components in various
<br>> toolkits (Swing, JavaFX, SWT) for
<br>> displaying the data layer.
<br>>
<br>> With an appropriate "model" (in the MVC arrangement) that includes
<br>> listeners, an adaptor-style hookup to any of the GUI toolkits should
<br>> be easy.
</div><br>Sounds good. I suppose once this is all done, integrating it into
<br>Eclipse should be relatively straightforward. The other stuff like an SL
<br>editor or a node-based editor are independent from that anyway (and I
<br>suppose, at the beginning, it would also be fine to just use the C/C++
<br>editor from the CDT plugin to write shaders).
<br><br>> I'm into the JVM mostly for the available languages... I love being
<br>> able to switch between Java, Groovy, JRuby, Python and Scala. I know
<br><br>I didn't know there are so many languages that can be run on the JVM! I
<br>heard about Jython, but never really knew how it actually works (another
<br>thing to look into eventually. There are just too many interesting
<br>things around... ;) )
<br>Does that mean I could write applets or Eclipse plugins in one of the
<br>above languages instead of Java? (it's interesting how all the languages
<br>start interleaving. Recently, I found "Pyjamas" which is a port of
<br>Google's Web Toolkit which means you can implement web applications in
<br>Python just as if it was a desktop GUI application. Pyjamas then
<br>translates the Python code into a set of JavaScript/HTML files and you
<br>can run the application inside a browser. If the regression test tool
<br>ever gets an overhaul, this might be an option to consider... :) )
<br><br>> The java.net package is the place to start for socket listening:
<br>> <a href="http://java.sun.com/javase/6/docs/api/java/net/package-summary.html" target="_top" rel="nofollow">http://java.sun.com/javase/6/docs/api/java/net/package-summary.html</a><br>> Basically, you open a socket and end up with a java.io.Reader, which
<br>> you can treat just like a FileReader or whatever else you're familiar with:
<br>> <a href="http://java.sun.com/docs/books/tutorial/networking/sockets/readingWriting.html" target="_top" rel="nofollow">http://java.sun.com/docs/books/tutorial/networking/sockets/readingWriting.html</a><br>> That could then listen for output from the normal Aqsis socket writer
<br>> (see src/tools/piqsl). I think that the low-level Dspy*-to-socket
<br>> stuff still belongs in C/C++ however.
<br><br>Thanks for the pointers! I wasn't sure if Java would provide other
<br>options as well and I also wasn't aware (anymore?) that piqsl uses a
<br>socket-based approach using XML.
<br>I suppose if the piqsl display driver could be generalized a bit, so
<br>that it can also launch other tools and communicate with them then half
<br>of the problem is already solved.
<br><br>Cheers,
<br><br>- Matthias -
<br><br><br>------------------------------------------------------------------------------
<br>Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>trial. Simplify your report design, integration and deployment - and focus on
<br>what you do best, core application coding. Discover what's new with
<br>Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26486445&i=0" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-26442021Re: Compiling Aqsis on Snow Leopard (OS X 10.6.x)2009-11-20T03:27:18Z2009-11-20T03:27:18ZPaul Gregory-4<div class="gmail_quote">2009/11/20 Chris Foster <span dir="ltr"><<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26442021&i=0" target="_top" rel="nofollow">chris42f@...</a>></span><br><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
There's some docs from the early days of piqsl development here:<br>
<br>
<a href="http://wiki.aqsis.org/dev/eqshibit" target="_blank" rel="nofollow">http://wiki.aqsis.org/dev/eqshibit</a><br>
<br>
I think they're still current; Paul will know better than me. I'm tempted to<br>
make the XML format a bit more consistent here and there, but it does work so<br>
maybe it's best to leave it alone ;-)<br>
<div class="im"><br></div></blockquote><div><br>Yes, those docs are pretty much valid, although a little sparse. Couple of points to note:<br><br>1. The Open sample shows a number of Parameters, this isn't fixed, it passes any specified parameters from the Aqsis side of the interface using that syntax, so you'd need to iterate over each parameter, deal with it's type, and decide if you understand/need it using the name.<br>
2. The CDATA in the data message is base64 encoded.<br><br><br>Paul<br></div></div>
<br />------------------------------------------------------------------------------
<br>Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>trial. Simplify your report design, integration and deployment - and focus on
<br>what you do best, core application coding. Discover what's new with
<br>Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br />_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26442021&i=1" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-26440853Re: Compiling Aqsis on Snow Leopard (OS X 10.6.x)2009-11-20T02:00:03Z2009-11-20T02:00:03ZChris Foster-5On Fri, Nov 20, 2009 at 5:33 PM, Jonathan Merritt
<br><<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26440853&i=0" target="_top" rel="nofollow">merritt@...</a>> wrote:
<br>> That could then listen for output from the normal Aqsis socket writer (see
<br>> src/tools/piqsl). I think that the low-level Dspy*-to-socket stuff still
<br>> belongs in C/C++ however.
<br><br>Sounds right to me.
<br><br>> Two questions I do have:
<br>> 1. Is there a spec / reference for the XML format that piqsl uses for
<br>> socket communication? I can construct most of it from
<br>> src/tools/piqsl/piqsl.cpp#processMessage, but documentation would be good.
<br><br>There's some docs from the early days of piqsl development here:
<br><br><a href="http://wiki.aqsis.org/dev/eqshibit" target="_top" rel="nofollow">http://wiki.aqsis.org/dev/eqshibit</a><br><br>I think they're still current; Paul will know better than me. I'm tempted to
<br>make the XML format a bit more consistent here and there, but it does work so
<br>maybe it's best to leave it alone ;-)
<br><br>> 2. How easy is it to separate the socket writing stuff in piqsl from the
<br>> FLTK GUI?
<br><br>It should be very easy, in fact the code is already in separate directories in
<br>the new source layout. From memory, the only thing you will need to change
<br>will be the program which the piqsl display automatically runs. I'd be happy
<br>to accept a patch which adds a display parameter allowing arbitrary framebuffer
<br>programs to be invoked (if this can't be done already that is).
<br><br>You may also need to change the cmake setup so that piqsl display is built
<br>independently of whether the necessary libraries exist to build piqsl itself.
<br><br>~Chris
<br><br>------------------------------------------------------------------------------
<br>Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>trial. Simplify your report design, integration and deployment - and focus on
<br>what you do best, core application coding. Discover what's new with
<br>Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26440853&i=1" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-26439428Re: Compiling Aqsis on Snow Leopard (OS X 10.6.x)2009-11-19T23:33:30Z2009-11-19T23:33:30ZJonathan Merritt-2Hi Matthias,
<br><br>On 20/11/2009, at 7:59 AM, Matthias Baas wrote:
<div class='shrinkable-quote'><br>> Jonathan Merritt wrote:
<br>>> Thanks for the link. I'll check out OIIO and iv sometime next week.
<br>>> I have also been contemplating writing a JVM-based display. I think
<br>>> that the cross-platform benefits and the multiple GUI toolkit options
<br>>> (JavaFX, Swing, SWT, etc.) might make it worthwhile. I also have far
<br>>> more experience with the various JVM languages than with C/C++ these
<br>>> days. :-)
<br>>
<br>> Have you considered implementing this framebuffer as an Eclipse view? I
<br>> think together with an editor plugin that supports SL, this would be a
<br>> nice IDE for writing shaders (for example, the framebuffer view could
<br>> automatically update while you write the shader and there could be GUI
<br>> controls for tweaking the shader parameters, etc.)
</div><br>That sounds like a great idea! Using Eclipse or Netbeans as the IDE backbone of a node-based editor would also make a nice project one day... :-)
<br><br>I was hoping to do something more general initially... a layered approach:
<br><br> Piqsl XML-based socket communication
<br> |
<br> Data layer of interfaces containing
<br> bucket data, listeners that are
<br> notified when buckets become
<br> available, etc.
<br> |
<br> Display layer; components in various
<br> toolkits (Swing, JavaFX, SWT) for
<br> displaying the data layer.
<br><br>With an appropriate "model" (in the MVC arrangement) that includes listeners, an adaptor-style hookup to any of the GUI toolkits should be easy.
<br><br>> (Being able to write Eclipse plugins is one of the reasons I've actually
<br>> started looking into Java (Android is another one), but so far, I have
<br>> no idea how a RenderMan display device would be implemented in Java...)
<br><br>I'm into the JVM mostly for the available languages... I love being able to switch between Java, Groovy, JRuby, Python and Scala. I know some people hate the current "language explosion", but I think it's fantastic, especially in the way it's getting to the guts of various issues like static-vs-dynamic typing, functional-vs-imperative styles, etc., and the way all of those work in a real development environment.
<br><br>The java.net package is the place to start for socket listening:
<br> <a href="http://java.sun.com/javase/6/docs/api/java/net/package-summary.html" target="_top" rel="nofollow">http://java.sun.com/javase/6/docs/api/java/net/package-summary.html</a><br>Basically, you open a socket and end up with a java.io.Reader, which you can treat just like a FileReader or whatever else you're familiar with:
<br> <a href="http://java.sun.com/docs/books/tutorial/networking/sockets/readingWriting.html" target="_top" rel="nofollow">http://java.sun.com/docs/books/tutorial/networking/sockets/readingWriting.html</a><br>That could then listen for output from the normal Aqsis socket writer (see src/tools/piqsl). I think that the low-level Dspy*-to-socket stuff still belongs in C/C++ however.
<br><br>Two questions I do have:
<br> 1. Is there a spec / reference for the XML format that piqsl uses for socket communication? I can construct most of it from src/tools/piqsl/piqsl.cpp#processMessage, but documentation would be good.
<br> 2. How easy is it to separate the socket writing stuff in piqsl from the FLTK GUI? (Would it be just as easy for me to start from scratch with the Dspy* stuff?)
<br><br>Thanks,
<br><br>Jonathan.
<br><br><br>------------------------------------------------------------------------------
<br>Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>trial. Simplify your report design, integration and deployment - and focus on
<br>what you do best, core application coding. Discover what's new with
<br>Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26439428&i=0" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-26434022Re: Compiling Aqsis on Snow Leopard (OS X 10.6.x)2009-11-19T12:59:48Z2009-11-19T12:59:48ZMatthias Baas-3Jonathan Merritt wrote:
<br>> Thanks for the link. I'll check out OIIO and iv sometime next week.
<br>> I have also been contemplating writing a JVM-based display. I think
<br>> that the cross-platform benefits and the multiple GUI toolkit options
<br>> (JavaFX, Swing, SWT, etc.) might make it worthwhile. I also have far
<br>> more experience with the various JVM languages than with C/C++ these
<br>> days. :-)
<br><br>Have you considered implementing this framebuffer as an Eclipse view? I
<br>think together with an editor plugin that supports SL, this would be a
<br>nice IDE for writing shaders (for example, the framebuffer view could
<br>automatically update while you write the shader and there could be GUI
<br>controls for tweaking the shader parameters, etc.)
<br><br>(Being able to write Eclipse plugins is one of the reasons I've actually
<br>started looking into Java (Android is another one), but so far, I have
<br>no idea how a RenderMan display device would be implemented in Java...)
<br><br>- Matthias -
<br><br><br>------------------------------------------------------------------------------
<br>Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>trial. Simplify your report design, integration and deployment - and focus on
<br>what you do best, core application coding. Discover what's new with
<br>Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26434022&i=0" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-26426800Re: Compiling Aqsis on Snow Leopard (OS X 10.6.x)2009-11-19T06:00:06Z2009-11-19T06:00:06ZTobias Sauerwein-2I haven't upgraded to Leopard yet but will do that this weekend,
<br>probably tomorrow.
<br>After that I'll give it a go and let you know how it went.
<br><br>About using a qt-based viewer: while qt looks nice the framework is
<br>quite heavy, so I'd be cautious, but then you only mentioned it as a
<br>possible alternative. I wonder if switching to fltk2 would be
<br>difficult and much work, presumed it compiles as 64bit. But also if it
<br>is already stable enough and would offer any benefit over the fltk
<br>version we're currently using.
<br><br>Cheers,
<br> Tobias
<br><br>On Thu, Nov 19, 2009 at 12:45 PM, Jonathan Merritt
<br><<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26426800&i=0" target="_top" rel="nofollow">merritt@...</a>> wrote:
<div class='shrinkable-quote'><br>> Hi Chris,
<br>>
<br>> On 19/11/2009, at 10:18 PM, Chris Foster wrote:
<br>>> On Thu, Nov 19, 2009 at 7:47 PM, Jonathan Merritt
<br>>> <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26426800&i=1" target="_top" rel="nofollow">merritt@...</a>> wrote:
<br>>>
<br>>> Sorry I can't really help with this Jonathan, just a minor off topic note...
<br>>>
<br>>>> c) obtaining an alternate framebuffer display driver; maybe writing my own (!)
<br>>>
<br>>> There's been some speculation that the image viewer (iv) from OIIO will be
<br>>> controllable by a socket interface sometime in the future and would be a
<br>>> possible alternative to piqsl (it uses qt).
<br>>>
<br>>> You're probably after something quick & simple, but on the off chance that
<br>>> you're interested I think that extending iv in this way would be a highly
<br>>> worthwhile project.
<br>>
<br>> Thanks for the link. I'll check out OIIO and iv sometime next week. I have also been contemplating writing a JVM-based display. I think that the cross-platform benefits and the multiple GUI toolkit options (JavaFX, Swing, SWT, etc.) might make it worthwhile. I also have far more experience with the various JVM languages than with C/C++ these days. :-)
<br>>
<br>> Jonathan.
<br>>
<br>>
<br>> ------------------------------------------------------------------------------
<br>> Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>> trial. Simplify your report design, integration and deployment - and focus on
<br>> what you do best, core application coding. Discover what's new with
<br>> Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>> _______________________________________________
<br>> Aqsis-development mailing list
<br>> <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26426800&i=2" target="_top" rel="nofollow">Aqsis-development@...</a>
<br>> <a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br>>
</div><br>------------------------------------------------------------------------------
<br>Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>trial. Simplify your report design, integration and deployment - and focus on
<br>what you do best, core application coding. Discover what's new with
<br>Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26426800&i=3" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-26424830Re: Compiling Aqsis on Snow Leopard (OS X 10.6.x)2009-11-19T03:45:22Z2009-11-19T03:45:22ZJonathan Merritt-2Hi Chris,
<br><br>On 19/11/2009, at 10:18 PM, Chris Foster wrote:
<div class='shrinkable-quote'><br>> On Thu, Nov 19, 2009 at 7:47 PM, Jonathan Merritt
<br>> <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26424830&i=0" target="_top" rel="nofollow">merritt@...</a>> wrote:
<br>>
<br>> Sorry I can't really help with this Jonathan, just a minor off topic note...
<br>>
<br>>> c) obtaining an alternate framebuffer display driver; maybe writing my own (!)
<br>>
<br>> There's been some speculation that the image viewer (iv) from OIIO will be
<br>> controllable by a socket interface sometime in the future and would be a
<br>> possible alternative to piqsl (it uses qt).
<br>>
<br>> You're probably after something quick & simple, but on the off chance that
<br>> you're interested I think that extending iv in this way would be a highly
<br>> worthwhile project.
</div><br>Thanks for the link. I'll check out OIIO and iv sometime next week. I have also been contemplating writing a JVM-based display. I think that the cross-platform benefits and the multiple GUI toolkit options (JavaFX, Swing, SWT, etc.) might make it worthwhile. I also have far more experience with the various JVM languages than with C/C++ these days. :-)
<br><br>Jonathan.
<br><br><br>------------------------------------------------------------------------------
<br>Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>trial. Simplify your report design, integration and deployment - and focus on
<br>what you do best, core application coding. Discover what's new with
<br>Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26424830&i=1" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-26424548Re: Compiling Aqsis on Snow Leopard (OS X 10.6.x)2009-11-19T03:18:17Z2009-11-19T03:18:17ZChris Foster-5On Thu, Nov 19, 2009 at 7:47 PM, Jonathan Merritt
<br><<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26424548&i=0" target="_top" rel="nofollow">merritt@...</a>> wrote:
<br><br>Sorry I can't really help with this Jonathan, just a minor off topic note...
<br><br>> c) obtaining an alternate framebuffer display driver; maybe writing my own (!)
<br><br>There's been some speculation that the image viewer (iv) from OIIO will be
<br>controllable by a socket interface sometime in the future and would be a
<br>possible alternative to piqsl (it uses qt).
<br><br>You're probably after something quick & simple, but on the off chance that
<br>you're interested I think that extending iv in this way would be a highly
<br>worthwhile project.
<br><br>~Chris
<br><br>------------------------------------------------------------------------------
<br>Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>trial. Simplify your report design, integration and deployment - and focus on
<br>what you do best, core application coding. Discover what's new with
<br>Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26424548&i=1" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-26423440Re: Compiling Aqsis on Snow Leopard (OS X 10.6.x)2009-11-19T01:47:10Z2009-11-19T01:47:10ZJonathan Merritt-2Hi Tobias (+ List :-),
<br><br>Have you had any success with a Snow Leopard compile?
<br><br>Just an update on my progress... I've been able to compile Aqsis without FLTK (actually as a 64-bit application), and it works fine without the framebuffer display driver. However, I *haven't* yet gone through the process of either:
<br> a) compiling FLTK with a 64-bit Quartz back-end, or
<br> b) compiling all Aqsis libs as 32-bit so they cooperate with FLTK, or
<br> c) obtaining an alternate framebuffer display driver; maybe writing my own (!)
<br>Option a) would be the best for me, but my naive efforts at compiling FLTK (version 1.3.x-r6916) as 64-bit have failed. I'm not even 100% sure it's possible.
<br><br>I may be able to spend some more time tinkering with this next week...
<br><br>Jonathan.
<br><br>On 06/11/2009, at 6:00 PM, Tobias Sauerwein wrote:
<div class='shrinkable-quote'><br>> Hi Jonathan,
<br>>
<br>> even if I'm not building under Snow Leopard yet, and despite the fact
<br>> that I have been very pro MacPorts in the past I switched to "build
<br>> everything by hand" because its easier to get a propper universal
<br>> binary than forcing MacPorts to do the right thing there.
<br>> But I know that Leon uses mainly MacPorts for our official build, so
<br>> he certainly has some tips for you.
<br>> Writing that up on the wiki would be great for other people who are
<br>> having the same issues. Saying that, I'll have to upgrade to SL
<br>> anyway, maybe today, so I might give it a chance this weekend.
<br>>
<br>> Cheers,
<br>> Tobias
<br>>
<br>> On Fri, Nov 6, 2009 at 2:55 AM, Jonathan Merritt <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26423440&i=0" target="_top" rel="nofollow">merritt@...</a>> wrote:
<br>>> Hi Everyone,
<br>>>
<br>>> Just wondering: is anyone currently compiling Aqsis under Snow
<br>>> Leopard? If so, are there any wiki instructions that I could follow?
<br>>>
<br>>> I've just started the process, using MacPorts (instead of hand-
<br>>> compiling everything as I have done in the past). However, it seems
<br>>> there is a problem with FLTK. AFAICT, MacPorts is compiling
<br>>> everything as 64-bit by default (which is also the Snow Leopard
<br>>> default), but FLTK can only be compiled in 32-bit mode because it
<br>>> still uses Carbon as the OS X interface layer. As a result, I get
<br>>> "file is not of the required architecture" errors with FLTK if I use
<br>>> the default CMake setting (x86_64), or with Boost and other libraries
<br>>> if I try to force i386. Does someone already have a compilation
<br>>> recipe that works? Or some appropriate kung-fu to make MacPorts
<br>>> behave? :-)
<br>>>
<br>>> It seems as though I need to make all of the libraries 32-bit. Is
<br>>> there some way to make MacPorts generate both 32-bit and 64-bit
<br>>> versions of libraries? And if so, how does that work out at run-time?
<br>>>
<br>>> Thanks,
<br>>>
<br>>> Jonathan Merritt.
</div><br><br>------------------------------------------------------------------------------
<br>Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>trial. Simplify your report design, integration and deployment - and focus on
<br>what you do best, core application coding. Discover what's new with
<br>Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26423440&i=1" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-26369891Re: aqsl on OSX 10.42009-11-16T02:46:52Z2009-11-16T02:46:52ZALoopingIconI have exactly the same problem, but with OSX 10.5.8 (plain leopard)
<br>renderer is ok, but compiling a shader causes
<br><br>dyld: Symbol not found: __ZNKSt9bad_alloc4whatEv
<br> Referenced from: /Applications/Aqsis.app/Contents/Resources/bin/../../Frameworks/libboost_wave.dylib
<br> Expected in: /usr/lib/libstdc++.6.dylib
<br><br>thanks in advance....
<br><br>p.
<br><br><blockquote class="quote light-black dark-border-color"><div class="quote light-border-color">
<div class="quote-author" style="font-weight: bold;">Matthias Baas-3 wrote:</div>
<div class="quote-message shrinkable-quote">Hi,
<br><br>when I tested Aqsis 1.6 on OSX 10.4 so far, I was only looking at the
<br>renderer itself which worked without any problem at all.
<br>But there is a slight problem with aqsl:
<br><br>$ aqsl testShader.sl
<br>dyld: Symbol not found: __ZNKSt9bad_alloc4whatEv
<br> Referenced from:
<br>/Applications/Graphics/Aqsis.app/Contents/Resources/bin/../../Frameworks/libboost_wave.dylib
<br> Expected in: /usr/lib/libstdc++.6.dylib
<br><br>Trace/BPT trap
<br><br>The missing symbol is this:
<br><br>$ c++filt
<br>__ZNKSt9bad_alloc4whatEv
<br>std::bad_alloc::what() const
<br><br>Looking at the libraries:
<br><br>$ otool -L
<br>/Applications/Graphics/Aqsis.app/Contents/Frameworks/libboost_wave.dylib
<br>/Applications/Graphics/Aqsis.app/Contents/Frameworks/libboost_wave.dylib:
<br> @executable_path/../../Frameworks/libboost_wave.dylib
<br>(compatibility version 0.0.0, current version 0.0.0)
<br> @executable_path/../../Frameworks/libboost_filesystem.dylib
<br>(compatibility version 0.0.0, current version 0.0.0)
<br> @executable_path/../../Frameworks/libboost_date_time.dylib
<br>(compatibility version 0.0.0, current version 0.0.0)
<br> @executable_path/../../Frameworks/libboost_thread.dylib
<br>(compatibility version 0.0.0, current version 0.0.0)
<br> /usr/lib/libstdc++.6.dylib (compatibility version 7.0.0, current
<br>version 7.9.0)
<br> /usr/lib/libgcc_s.1.dylib (compatibility version 1.0.0, current
<br>version 103.0.0)
<br> /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current
<br>version 124.1.1)
<br><br>$ otool -L /usr/lib/libstdc++.6.dylib
<br>/usr/lib/libstdc++.6.dylib:
<br> /usr/lib/libstdc++.6.dylib (compatibility version 7.0.0, current
<br>version 7.4.0)
<br> /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current
<br>version 88.3.7)
<br> /usr/lib/libgcc_s.1.dylib (compatibility version 1.0.0, current
<br>version 1.0.0)
<br><br><br>So does that mean aqsl was compiled on a machine that had libstdc++
<br>v7.9.0 whereas I have v7.4.0? But don't they both claim to be compatible
<br>with v7.0.0 so that this unresolved symbol shouldn't have happened? (or
<br>was bad_alloc::what() added after 7.4? But then shouldn't it be recorded
<br>in libboost_wave somewhere that it needs a higher version of libstdc++?)
<br><br>- Matthias -
<br><br><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry(R) Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9 - 12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconference" target="_top" rel="nofollow">http://p.sf.net/sfu/devconference</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br>Aqsis-development@lists.sourceforge.net
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a></div>
</div></blockquote>
<p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-26227688Re: Compiling Aqsis on Snow Leopard (OS X 10.6.x)2009-11-05T23:00:58Z2009-11-05T23:00:58ZTobias Sauerwein-2Hi Jonathan,
<br><br>even if I'm not building under Snow Leopard yet, and despite the fact
<br>that I have been very pro MacPorts in the past I switched to "build
<br>everything by hand" because its easier to get a propper universal
<br>binary than forcing MacPorts to do the right thing there.
<br>But I know that Leon uses mainly MacPorts for our official build, so
<br>he certainly has some tips for you.
<br>Writing that up on the wiki would be great for other people who are
<br>having the same issues. Saying that, I'll have to upgrade to SL
<br>anyway, maybe today, so I might give it a chance this weekend.
<br><br>Cheers,
<br> Tobias
<br><br><br><br><br><br>On Fri, Nov 6, 2009 at 2:55 AM, Jonathan Merritt <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26227688&i=0" target="_top" rel="nofollow">merritt@...</a>> wrote:
<div class='shrinkable-quote'><br>> Hi Everyone,
<br>>
<br>> Just wondering: is anyone currently compiling Aqsis under Snow
<br>> Leopard? If so, are there any wiki instructions that I could follow?
<br>>
<br>> I've just started the process, using MacPorts (instead of hand-
<br>> compiling everything as I have done in the past). However, it seems
<br>> there is a problem with FLTK. AFAICT, MacPorts is compiling
<br>> everything as 64-bit by default (which is also the Snow Leopard
<br>> default), but FLTK can only be compiled in 32-bit mode because it
<br>> still uses Carbon as the OS X interface layer. As a result, I get
<br>> "file is not of the required architecture" errors with FLTK if I use
<br>> the default CMake setting (x86_64), or with Boost and other libraries
<br>> if I try to force i386. Does someone already have a compilation
<br>> recipe that works? Or some appropriate kung-fu to make MacPorts
<br>> behave? :-)
<br>>
<br>> It seems as though I need to make all of the libraries 32-bit. Is
<br>> there some way to make MacPorts generate both 32-bit and 64-bit
<br>> versions of libraries? And if so, how does that work out at run-time?
<br>>
<br>> Thanks,
<br>>
<br>> Jonathan Merritt.
<br>>
<br>>
<br>> ------------------------------------------------------------------------------
<br>> Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>> trial. Simplify your report design, integration and deployment - and focus on
<br>> what you do best, core application coding. Discover what's new with
<br>> Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>> _______________________________________________
<br>> Aqsis-development mailing list
<br>> <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26227688&i=1" target="_top" rel="nofollow">Aqsis-development@...</a>
<br>> <a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br>>
</div><br>------------------------------------------------------------------------------
<br>Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>trial. Simplify your report design, integration and deployment - and focus on
<br>what you do best, core application coding. Discover what's new with
<br>Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26227688&i=2" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-26225851Compiling Aqsis on Snow Leopard (OS X 10.6.x)2009-11-05T17:55:37Z2009-11-05T17:55:37ZJonathan Merritt-2Hi Everyone,
<br><br>Just wondering: is anyone currently compiling Aqsis under Snow
<br>Leopard? If so, are there any wiki instructions that I could follow?
<br><br>I've just started the process, using MacPorts (instead of hand-
<br>compiling everything as I have done in the past). However, it seems
<br>there is a problem with FLTK. AFAICT, MacPorts is compiling
<br>everything as 64-bit by default (which is also the Snow Leopard
<br>default), but FLTK can only be compiled in 32-bit mode because it
<br>still uses Carbon as the OS X interface layer. As a result, I get
<br>"file is not of the required architecture" errors with FLTK if I use
<br>the default CMake setting (x86_64), or with Boost and other libraries
<br>if I try to force i386. Does someone already have a compilation
<br>recipe that works? Or some appropriate kung-fu to make MacPorts
<br>behave? :-)
<br><br>It seems as though I need to make all of the libraries 32-bit. Is
<br>there some way to make MacPorts generate both 32-bit and 64-bit
<br>versions of libraries? And if so, how does that work out at run-time?
<br><br>Thanks,
<br><br>Jonathan Merritt.
<br><br><br>------------------------------------------------------------------------------
<br>Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day
<br>trial. Simplify your report design, integration and deployment - and focus on
<br>what you do best, core application coding. Discover what's new with
<br>Crystal Reports now. <a href="http://p.sf.net/sfu/bobj-july" target="_top" rel="nofollow">http://p.sf.net/sfu/bobj-july</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=26225851&i=0" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25948604aqsl on OSX 10.42009-10-18T10:22:49Z2009-10-18T10:22:49ZMatthias Baas-3Hi,
<br><br>when I tested Aqsis 1.6 on OSX 10.4 so far, I was only looking at the
<br>renderer itself which worked without any problem at all.
<br>But there is a slight problem with aqsl:
<br><br>$ aqsl testShader.sl
<br>dyld: Symbol not found: __ZNKSt9bad_alloc4whatEv
<br> Referenced from:
<br>/Applications/Graphics/Aqsis.app/Contents/Resources/bin/../../Frameworks/libboost_wave.dylib
<br> Expected in: /usr/lib/libstdc++.6.dylib
<br><br>Trace/BPT trap
<br><br>The missing symbol is this:
<br><br>$ c++filt
<br>__ZNKSt9bad_alloc4whatEv
<br>std::bad_alloc::what() const
<br><br>Looking at the libraries:
<br><br>$ otool -L
<br>/Applications/Graphics/Aqsis.app/Contents/Frameworks/libboost_wave.dylib
<br>/Applications/Graphics/Aqsis.app/Contents/Frameworks/libboost_wave.dylib:
<br> @executable_path/../../Frameworks/libboost_wave.dylib
<br>(compatibility version 0.0.0, current version 0.0.0)
<br> @executable_path/../../Frameworks/libboost_filesystem.dylib
<br>(compatibility version 0.0.0, current version 0.0.0)
<br> @executable_path/../../Frameworks/libboost_date_time.dylib
<br>(compatibility version 0.0.0, current version 0.0.0)
<br> @executable_path/../../Frameworks/libboost_thread.dylib
<br>(compatibility version 0.0.0, current version 0.0.0)
<br> /usr/lib/libstdc++.6.dylib (compatibility version 7.0.0, current
<br>version 7.9.0)
<br> /usr/lib/libgcc_s.1.dylib (compatibility version 1.0.0, current
<br>version 103.0.0)
<br> /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current
<br>version 124.1.1)
<br><br>$ otool -L /usr/lib/libstdc++.6.dylib
<br>/usr/lib/libstdc++.6.dylib:
<br> /usr/lib/libstdc++.6.dylib (compatibility version 7.0.0, current
<br>version 7.4.0)
<br> /usr/lib/libSystem.B.dylib (compatibility version 1.0.0, current
<br>version 88.3.7)
<br> /usr/lib/libgcc_s.1.dylib (compatibility version 1.0.0, current
<br>version 1.0.0)
<br><br><br>So does that mean aqsl was compiled on a machine that had libstdc++
<br>v7.9.0 whereas I have v7.4.0? But don't they both claim to be compatible
<br>with v7.0.0 so that this unresolved symbol shouldn't have happened? (or
<br>was bad_alloc::what() added after 7.4? But then shouldn't it be recorded
<br>in libboost_wave somewhere that it needs a higher version of libstdc++?)
<br><br>- Matthias -
<br><br><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry(R) Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9 - 12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconference" target="_top" rel="nofollow">http://p.sf.net/sfu/devconference</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25948604&i=0" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25764402Re: [Aqsis-commits] [SCM] Aqsis Renderer branch, master, updated. Release_1.6.0_Phase2-69-g6b970b02009-10-06T01:02:01Z2009-10-06T01:02:01ZChris Foster-5On Tue, Oct 6, 2009 at 2:43 AM, renderguy
<br><<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25764402&i=0" target="_top" rel="nofollow">ltatkinson@...</a>> wrote:
<br>> - Log -----------------------------------------------------------------
<br>> commit 6b970b0e3d82aebbfd0badf8a95f4627251e61db
<br>> Author: Leon Atkinson <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25764402&i=1" target="_top" rel="nofollow">latkinson@...</a>>
<br>> Date: Mon Oct 5 16:42:41 2009 +0000
<br>>
<br>> Missed some file from previous 'Neqsus' commit
<br><br>This commit has lots of files with windows file endings. Maybe this is worth
<br>fixing before the official release of the 1.6 source?
<br><br>Just a heads-up, given that I noticed 1.6 isn't quite out yet ;-)
<br><br>~Chris
<br><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25764402&i=2" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25751785Re: patch for bug id 2813534 "Escaped strings in RSL not working"2009-10-05T07:24:58Z2009-10-05T07:24:58ZChris Foster-5On Tue, Oct 6, 2009 at 12:05 AM, Ricardo Mayor <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25751785&i=0" target="_top" rel="nofollow">ricardo.mayor@...</a>> wrote:
<br>> My first option was to return the interpreted character but this assumes
<br>> that you always have a char to return. This is not the case when the user
<br>> writes incorrect numeric escapes sequences. For example:
<br>> "\xp" //Hex incorrect number
<br>> "\0AF"//Octal incorrect number
<br><br>Fair point. (Though I think the octal example is valid... it's \0 followed by
<br>the letters AF which are unrelated.)
<br><br>[...]
<br><br>>> I wonder whether this is the right thing to do? Maybe the user should
<br>>> be allowed to insert NULL into the string if they want? Mind you, it's
<br>>> probably going to cause problems if we use char* instead of std::string
<br>>> anywhere since then the string will be prematurely terminated...
<br>>>
<br>> In this case the null is not allowed. If the user create a strint "\000",
<br>> this will not be accepted.
<br><br>Why can't it be accepted though? It is legal in C, and we're trying to follow
<br>the C rules because the RISpec says so. In the following C++ code, it's
<br>perfectly legal to use \0 in a string, and it gets interpreted as 0:
<br><br>----------
<br>#include <iostream>
<br>#include <string>
<br><br>int main()
<br>{
<br> const char s[] = "asdf\0asdf";
<br> std::string str(s, s+sizeof(s));
<br><br> std::cout << str << "\n";
<br><br> std::cout << "str[4] = " << int(str[4]) << "\n";
<br><br> return 0;
<br>}
<br>----------
<br><br><br>~Chris.
<br><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25751785&i=1" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25751481Re: patch for bug id 2813534 "Escaped strings in RSL not working"2009-10-05T07:05:20Z2009-10-05T07:05:20ZRicardo MayorI agree with all the changes you suggested except with two.<br><br><div class="gmail_quote">On Mon, Oct 5, 2009 at 1:49 PM, Chris Foster <span dir="ltr"><<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25751481&i=0" target="_top" rel="nofollow">chris42f@...</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><br>
This function signature would be a lot cleaner if you avoid passing in c<br>
(just unget() it) and if you return the interpreted character rather<br>
than appending it to the string s.<br></blockquote><div><br>My first option was to return the interpreted character but this assumes that you always have a char to return. This is not the case when the user writes incorrect numeric escapes sequences. For example:<br>
"\xp" //Hex incorrect number<br>"\0AF"//Octal incorrect number<br><br>In both cases the numeric sequence is invalid and the actual code does not add any char to the string. If the function returns a char, it doesn't have a char to return. That's the reason why I pass the CqString, to encapsulate this functionality inside the function. In the current examples, the final string are:<br>
"p"<br>"AF"<br> </div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
> +{<br>
> + CqString a("");<br>
> + bool isHexadecimal = (c =='x');<br>
> + unsigned int maxLength;<br>
> +<br>
> + if (isHexadecimal)<br>
> + {<br>
> + maxLength = 2;<br>
> + }<br>
> + else<br>
> + {<br>
> + a += c;<br>
> + maxLength = 3;<br>
> + }<br>
> + c = pFile->get();<br>
> +<br>
> + while ((( c >= '0' && c <= '9') || ((( c >= 'a' && c <= 'f') || ( c >= 'A' && c <= 'F')) && isHexadecimal)) && (a.length() <maxLength))<br>
> + {<br>
> + a += c;<br>
> + c = pFile->get();<br>
> + }<br>
> +<br>
> + int numericalBase;<br>
> +<br>
> + if(isHexadecimal)<br>
> + numericalBase = 16;<br>
> + else<br>
> + numericalBase = 8;<br>
> + char result = strtoul(a.c_str(), NULL, numericalBase);<br>
> +<br>
> + if (result != 0) s += result;<br>
<br></blockquote><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
I wonder whether this is the right thing to do? Maybe the user should<br>
be allowed to insert NULL into the string if they want? Mind you, it's<br>
probably going to cause problems if we use char* instead of std::string<br>
anywhere since then the string will be prematurely terminated...<br>
<br></blockquote><div>In this case the null is not allowed. If the user create a strint "\000", this will not be accepted. <br></div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
> + pFile->unget();<br>
> +}<br>
><br></blockquote></div><br>
<br />------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br />_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25751481&i=1" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25750261Re: patch for bug id 2813534 "Escaped strings in RSL not working"2009-10-05T05:49:52Z2009-10-05T05:49:52ZChris Foster-5Hi Ricardo,
<br><br>On Sun, Oct 04, 2009 at 08:43:11PM +0100, Ricardo Mayor wrote:
<br>> Please find attached another version of the patch.
<br><br>Thanks!
<br><br>There's still a few issues, but the patch certainly looks better now.
<br>Since we're just about to release aqsis-1.6 I'm going to commit the
<br>patch anyway, and clean up the few bits I think are most important
<br>myself.
<br><br>I'm quite picky with code quality, so bear with me ;-) Here's a review
<br>of the patch. The main things which I think are fairly important and
<br>I'll fix myself (only because of the tight shedule before the release)
<br>are:
<br><br>1) Some inaccurate function docs in shadervm.h
<br>2) Declare GetString() and GetNumericEscapeChar() as static since they
<br> don't access any CqShaderVM member data
<br><br>When checking the patch I also noticed that the function SO_sprintf()
<br>(for printing strings from a shader program) actually translates escape
<br>characters itself. It uses the function CqString::TranslateEscapes()
<br>which I didn't know about. With the new code the translation has
<br>already happened when reading the shader program, so I'll fix that.
<br><br><div class='shrinkable-quote'><br>> diff --git a/libs/shadervm/shadervm.cpp b/libs/shadervm/shadervm.cpp
<br>> index b498adf..5861646 100644
<br>> --- a/libs/shadervm/shadervm.cpp
<br>> +++ b/libs/shadervm/shadervm.cpp
<br>> @@ -1341,6 +1334,127 @@ void CqShaderVM::LoadProgram( std::istream* pFile )
<br>> }
<br>> }
<br>>
<br>> +CqString CqShaderVM::GetString(std::istream* pFile)
<br>> +{
<br>> + ( *pFile ) >> std::ws;
<br>> + char c;
<br>> + CqString s( "" );
<br>> + bool escapeChar = false;
<br>> +
<br>> + pFile->get();
<br>> + while ( (( c = pFile->get() ) != '"') || escapeChar )
<br>> + {
<br>> + if(escapeChar)
</div><br>The presence of the escapeChar variable seems to complicate things
<br>a bit unnecessarily. It would be simpler if you allow yourself
<br>more than one place to call pFile->get(). In particular, you can have
<br><br>if(c == '\\')
<br>{
<br> c = pFile->get();
<br> switch(c)
<br> {
<br> //...
<br> }
<br>}
<br>else
<br> s += c;
<br><br>which avoids the escapeChar variable completely. Other than that, this
<br>function looks about right - you seem to have all the right escape
<br>characters there.
<br><div class='shrinkable-quote'><br>> + {
<br>> + //Treatment for escape char
<br>> + switch(c)
<br>> + {
<br>> + case 'a'://Bell (alert)
<br>> + s += '\a';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case 'v'://Vertical tab
<br>> + s += '\v';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case '\''://Single quotation mark
<br>> + s += "'";
<br>> + escapeChar = false;
<br>> + break;
<br>> + case '?'://Literal question mark
<br>> + s += '\?';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case 'n'://New line
<br>> + s += '\n';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case 'r'://Carriage return
<br>> + s += '\r';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case 't'://Horizontal tab
<br>> + s += '\t';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case 'b'://Backspace
<br>> + s += '\b';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case 'f'://Formfeed
<br>> + s += '\f';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case '"'://Double quotation mark
<br>> + s += '"';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case '\\'://Backslash
<br>> + s += '\\';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case 'x'://Hexadecimal
<br>> + case '0'://octal
<br>> + case '1':
<br>> + case '2':
<br>> + case '3':
<br>> + case '4':
<br>> + case '5':
<br>> + case '6':
<br>> + case '7':
<br>> + case '8':
<br>> + case '9':
<br>> + GetNumericEscapeChar(pFile, s, c);
<br>> + escapeChar = false;
<br>> + break;
<br>> + default :
<br>> + escapeChar = false;
<br>> + break;
<br>> + }
<br>> + }
<br>> + else
<br>> + if (c == '\\')
<br>> + escapeChar = true;
<br>> + else
<br>> + s += c;
<br>> + }
<br>> + return s;
<br>> +}
<br>>
<br>> +void CqShaderVM::GetNumericEscapeChar(std::istream* pFile, CqString &s, char c)
</div><br>This function signature would be a lot cleaner if you avoid passing in c
<br>(just unget() it) and if you return the interpreted character rather
<br>than appending it to the string s.
<br><div class='shrinkable-quote'><br>> +{
<br>> + CqString a("");
<br>> + bool isHexadecimal = (c =='x');
<br>> + unsigned int maxLength;
<br>> +
<br>> + if (isHexadecimal)
<br>> + {
<br>> + maxLength = 2;
<br>> + }
<br>> + else
<br>> + {
<br>> + a += c;
<br>> + maxLength = 3;
<br>> + }
<br>> + c = pFile->get();
<br>> +
<br>> + while ((( c >= '0' && c <= '9') || ((( c >= 'a' && c <= 'f') || ( c >= 'A' && c <= 'F')) && isHexadecimal)) && (a.length() <maxLength))
<br>> + {
<br>> + a += c;
<br>> + c = pFile->get();
<br>> + }
<br>> +
<br>> + int numericalBase;
<br>> +
<br>> + if(isHexadecimal)
<br>> + numericalBase = 16;
<br>> + else
<br>> + numericalBase = 8;
<br>> + char result = strtoul(a.c_str(), NULL, numericalBase);
<br>> +
<br>> + if (result != 0) s += result;
</div><br>I wonder whether this is the right thing to do? Maybe the user should
<br>be allowed to insert NULL into the string if they want? Mind you, it's
<br>probably going to cause problems if we use char* instead of std::string
<br>anywhere since then the string will be prematurely terminated...
<br><div class='shrinkable-quote'><br>> + pFile->unget();
<br>> +}
<br>>
<br>> //---------------------------------------------------------------------
<br>> /** Ready the shader for execution.
<br>> diff --git a/libs/shadervm/shadervm.h b/libs/shadervm/shadervm.h
<br>> index 11eb227..9e55f90 100644
<br>> --- a/libs/shadervm/shadervm.h
<br>> +++ b/libs/shadervm/shadervm.h
<br>> @@ -246,6 +246,19 @@ class AQSIS_SHADERVM_SHARE CqShaderVM : public CqShaderStack, public IqShader, p
<br>> bool m_outsideWorld; ///< Flag indicating this shader was declared outside the world.
<br>> IqRenderer* m_pRenderContext;
<br>>
<br>> +
<br>> + /** Get the char defined in the Numeric Escape char
<br>> + * \param pFile shader stream
<br>> + * \return String
<br>> + */
<br>> + CqString GetString(std::istream* pFile);
</div><br>The documentation is confusing here.
<br><br>Also, note that GetString doesn't modify or even reference the state of
<br>CqShaderVM. Therefore, it should be a static method:
<br><br> static CqString GetString(std::istream* pFile);
<br><br>(At the very least it should be const method:
<br> CqString GetString(std::istream* pFile) const;
<br>)
<br><br>> + /** Get the char defined in the Numeric Escape char
<br>> + * \param pFile shader stream
<br>> + * \param s result string
<br>> + * \param c, current character
<br>> + * \return char equivalence for the numeric escape sequence
<br>> + */
<br>> + void GetNumericEscapeChar(std::istream* pFile, CqString &s, char c) ;
<br><br>Similar thing here. The docs seem out of date, and this function should
<br>be static too.
<br><br>That's it, thanks for the patch - it's committed now along with the
<br>minor changes by me in a separate commit.
<br>~Chris.
<br><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25750261&i=0" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25741399Re: patch for bug id 2813534 "Escaped strings in RSL not working"2009-10-04T12:43:11Z2009-10-04T12:43:11ZRicardo MayorPlease find attached another version of the patch. I hope this version fulfill all the discussed requirements<br><br>Best regards<br><br><div class="gmail_quote">On Thu, Oct 1, 2009 at 12:36 PM, Chris Foster <span dir="ltr"><<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25741399&i=0" target="_top" rel="nofollow">chris42f@...</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;"><div class="im">On Thu, Oct 1, 2009 at 6:10 PM, Ricardo Mayor <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25741399&i=1" target="_top" rel="nofollow">ricardo.mayor@...</a>> wrote:<br>
> Hi Chris,<br>
> I'll implement all your suggestion and corrections in a new patch. Most of<br>
> them are easy to add I think. Find some comments in the email<br>
<br>
</div>Cool, thanks.<br>
<div class="im"><br>
>> Please use opening braces on their own line for consistency with the rest<br>
>> of<br>
>> the codebase.<br>
><br>
> I installed Artistic style to check the code style but.. Where is the<br>
> style file?<br>
<br>
</div>I personally dislike automatic code formatting because it's possible to do a<br>
better job by hand if you're careful. The most important thing to keep in<br>
mind is to follow the style of the code around that which you're fixing to<br>
keep things consistent.<br>
<br>
Unfortunately there is more than one styles used in the aqsis codebase (eg,<br>
capitalization of class member functions is inconsistent). The current<br>
"standard" such as it is can be found in<br>
<br>
doc/coding_standard/codestyle.*<br>
<br>
You will also find the astylerc file in doc/coding_standard, thought as I<br>
said, I prefer to do careful manual formatting.<br>
<div class="im"><br>
>> When I was reading through this code, I got rather confused by the<br>
>> handling of<br>
>> capturedNumber, since it appears in two different places and there's two<br>
>> different places where it's converted to a char using strtoul(). I<br>
>> suggest<br>
>> just using a nested loop inside this part of the case statement to confine<br>
>> octal handling logic to a single place in the code. When you encounter<br>
>> the<br>
>> end of the octal number you can use the function pFile->unget() to put<br>
>> back<br>
>> the terminating character into the input stream. (For an example of<br>
>> something<br>
>> very similar, see CqRibLexer::readString() in riblexer.cpp.)<br>
><br>
> I wrote the code in this way to avoid using file->unget() when the length of<br>
> the number is less than 3. Anyway I´ll check it<br>
<br>
</div>Yes, you definitely need to use unget() to write the code how I'm suggesting,<br>
but I see no reason to avoid using it. I think it will make this code clearer<br>
and shorter ;-)<br>
<div><div></div><div class="h5"><br>
~Chris.<br>
<br>
------------------------------------------------------------------------------<br>
Come build with us! The BlackBerry&reg; Developer Conference in SF, CA<br>
is the only developer event you need to attend this year. Jumpstart your<br>
developing skills, take BlackBerry mobile applications to market and stay<br>
ahead of the curve. Join us from November 9&#45;12, 2009. Register now&#33;<br>
<a href="http://p.sf.net/sfu/devconf" target="_blank" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>
_______________________________________________<br>
Aqsis-development mailing list<br>
<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25741399&i=2" target="_top" rel="nofollow">Aqsis-development@...</a><br>
<a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_blank" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br>
</div></div></blockquote></div><br>
<br /><tt>[escape_char_v2.patch]</tt><br /><hr align="left" width="300" /><tt>diff --git a/libs/shadervm/shadervm.cpp b/libs/shadervm/shadervm.cpp
<br>index b498adf..5861646 100644
<br>--- a/libs/shadervm/shadervm.cpp
<br>+++ b/libs/shadervm/shadervm.cpp
<br>@@ -1276,14 +1276,7 @@ void CqShaderVM::LoadProgram( std::istream* pFile )
<br> break;
<br> case type_string:
<br> {
<br>- ( *pFile ) >> std::ws;
<br>- char c;
<br>- CqString s( "" );
<br>- pFile->get
<br>- ();
<br>- while ( ( c = pFile->get
<br>- () ) != '"' )
<br>- s += c;
<br>+ CqString s = GetString(pFile);
<br> AddString( s.c_str(), pProgramArea );
<br> }
<br> break;
<br>@@ -1341,6 +1334,127 @@ void CqShaderVM::LoadProgram( std::istream* pFile )
<br> }
<br> }
<br>
<br>+CqString CqShaderVM::GetString(std::istream* pFile)
<br>+{
<br>+ ( *pFile ) >> std::ws;
<br>+ char c;
<br>+ CqString s( "" );
<br>+ bool escapeChar = false;
<br>+
<br>+ pFile->get();
<br>+ while ( (( c = pFile->get() ) != '"') || escapeChar )
<br>+ {
<br>+ if(escapeChar)
<br>+ {
<br>+ //Treatment for escape char
<br>+ switch(c)
<br>+ {
<br>+ case 'a'://Bell (alert)
<br>+ s += '\a';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case 'v'://Vertical tab
<br>+ s += '\v';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case '\''://Single quotation mark
<br>+ s += "'";
<br>+ escapeChar = false;
<br>+ break;
<br>+ case '?'://Literal question mark
<br>+ s += '\?';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case 'n'://New line
<br>+ s += '\n';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case 'r'://Carriage return
<br>+ s += '\r';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case 't'://Horizontal tab
<br>+ s += '\t';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case 'b'://Backspace
<br>+ s += '\b';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case 'f'://Formfeed
<br>+ s += '\f';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case '"'://Double quotation mark
<br>+ s += '"';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case '\\'://Backslash
<br>+ s += '\\';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case 'x'://Hexadecimal
<br>+ case '0'://octal
<br>+ case '1':
<br>+ case '2':
<br>+ case '3':
<br>+ case '4':
<br>+ case '5':
<br>+ case '6':
<br>+ case '7':
<br>+ case '8':
<br>+ case '9':
<br>+ GetNumericEscapeChar(pFile, s, c);
<br>+ escapeChar = false;
<br>+ break;
<br>+ default :
<br>+ escapeChar = false;
<br>+ break;
<br>+ }
<br>+ }
<br>+ else
<br>+ if (c == '\\')
<br>+ escapeChar = true;
<br>+ else
<br>+ s += c;
<br>+ }
<br>+ return s;
<br>+}
<br>+
<br>+void CqShaderVM::GetNumericEscapeChar(std::istream* pFile, CqString &s, char c)
<br>+{
<br>+ CqString a("");
<br>+ bool isHexadecimal = (c =='x');
<br>+ unsigned int maxLength;
<br>+
<br>+ if (isHexadecimal)
<br>+ {
<br>+ maxLength = 2;
<br>+ }
<br>+ else
<br>+ {
<br>+ a += c;
<br>+ maxLength = 3;
<br>+ }
<br>+ c = pFile->get();
<br>+
<br>+ while ((( c >= '0' && c <= '9') || ((( c >= 'a' && c <= 'f') || ( c >= 'A' && c <= 'F')) && isHexadecimal)) && (a.length() <maxLength))
<br>+ {
<br>+ a += c;
<br>+ c = pFile->get();
<br>+ }
<br>+
<br>+ int numericalBase;
<br>+
<br>+ if(isHexadecimal)
<br>+ numericalBase = 16;
<br>+ else
<br>+ numericalBase = 8;
<br>+ char result = strtoul(a.c_str(), NULL, numericalBase);
<br>+
<br>+ if (result != 0) s += result;
<br>+ pFile->unget();
<br>+}
<br>
<br> //---------------------------------------------------------------------
<br> /** Ready the shader for execution.
<br>diff --git a/libs/shadervm/shadervm.h b/libs/shadervm/shadervm.h
<br>index 11eb227..9e55f90 100644
<br>--- a/libs/shadervm/shadervm.h
<br>+++ b/libs/shadervm/shadervm.h
<br>@@ -246,6 +246,19 @@ class AQSIS_SHADERVM_SHARE CqShaderVM : public CqShaderStack, public IqShader, p
<br> bool m_outsideWorld; ///< Flag indicating this shader was declared outside the world.
<br> IqRenderer* m_pRenderContext;
<br>
<br>+
<br>+ /** Get the char defined in the Numeric Escape char
<br>+ * \param pFile shader stream
<br>+ * \return String
<br>+ */
<br>+ CqString GetString(std::istream* pFile);
<br>+ /** Get the char defined in the Numeric Escape char
<br>+ * \param pFile shader stream
<br>+ * \param s result string
<br>+ * \param c, current character
<br>+ * \return char equivalence for the numeric escape sequence
<br>+ */
<br>+ void GetNumericEscapeChar(std::istream* pFile, CqString &s, char c) ;
<br> /** Determine whether the program execution has finished.
<br> */
<br> bool fDone()
<br></tt><hr align="left" width="300" /><br />------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br />_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25741399&i=3" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25697338Re: patch for bug id 2813534 "Escaped strings in RSL not working"2009-10-01T04:36:38Z2009-10-01T04:36:38ZChris Foster-5On Thu, Oct 1, 2009 at 6:10 PM, Ricardo Mayor <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25697338&i=0" target="_top" rel="nofollow">ricardo.mayor@...</a>> wrote:
<br>> Hi Chris,
<br>> I'll implement all your suggestion and corrections in a new patch. Most of
<br>> them are easy to add I think. Find some comments in the email
<br><br>Cool, thanks.
<br><br>>> Please use opening braces on their own line for consistency with the rest
<br>>> of
<br>>> the codebase.
<br>>
<br>> I installed Artistic style to check the code style but.. Where is the
<br>> style file?
<br><br>I personally dislike automatic code formatting because it's possible to do a
<br>better job by hand if you're careful. The most important thing to keep in
<br>mind is to follow the style of the code around that which you're fixing to
<br>keep things consistent.
<br><br>Unfortunately there is more than one styles used in the aqsis codebase (eg,
<br>capitalization of class member functions is inconsistent). The current
<br>"standard" such as it is can be found in
<br><br>doc/coding_standard/codestyle.*
<br><br>You will also find the astylerc file in doc/coding_standard, thought as I
<br>said, I prefer to do careful manual formatting.
<br><div class='shrinkable-quote'><br>>> When I was reading through this code, I got rather confused by the
<br>>> handling of
<br>>> capturedNumber, since it appears in two different places and there's two
<br>>> different places where it's converted to a char using strtoul(). I
<br>>> suggest
<br>>> just using a nested loop inside this part of the case statement to confine
<br>>> octal handling logic to a single place in the code. When you encounter
<br>>> the
<br>>> end of the octal number you can use the function pFile->unget() to put
<br>>> back
<br>>> the terminating character into the input stream. (For an example of
<br>>> something
<br>>> very similar, see CqRibLexer::readString() in riblexer.cpp.)
<br>>
<br>> I wrote the code in this way to avoid using file->unget() when the length of
<br>> the number is less than 3. Anyway I´ll check it
</div><br>Yes, you definitely need to use unget() to write the code how I'm suggesting,
<br>but I see no reason to avoid using it. I think it will make this code clearer
<br>and shorter ;-)
<br><br>~Chris.
<br><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25697338&i=1" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25694651Re: patch for bug id 2813534 "Escaped strings in RSL not working"2009-10-01T01:10:41Z2009-10-01T01:10:41ZRicardo MayorHi Chris,<br>I'll implement all your suggestion and corrections in a new patch. Most of them are easy to add I think. Find some comments in the email<br><br><br><div class="gmail_quote">On Wed, Sep 30, 2009 at 10:58 PM, Chris Foster <span dir="ltr"><<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25694651&i=0" target="_top" rel="nofollow">chris42f@...</a>></span> wrote:<br>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">Hi Ricardo,<br>
<div class="im"><br>
On Sun, Sep 27, 2009 at 04:12:57PM +0100, Ricardo Mayor wrote:<br>
> Hello,<br>
> Find attached a patch for the bug "*Escaped strings in RSL not working".<br>
> According *to ri specs, the supported escape string are<br>
><br>
> \n linefeed (newline)<br>
> \r carriage return<br>
> \t horizontal tab<br>
> \b backspace<br>
> \f form feed<br>
> \\ backslash<br>
> \" double quote<br>
> \ddd character code ddd (octal)<br>
> \newline no character ? both are ignored<br>
<br>
</div>I think these may be the escape sequences specified in the part of the RISpec<br>
which deals with the RIB format, not the renderman shading language?<br>
<br>
Here's what I see in my copy of the RISpec (around page 117):<br>
<br>
" Strings are used to name external objects (texture maps, for<br>
example). String literals (or<br>
" constants) are enclosed in double quotes, as in the C language, and<br>
may contain any of the<br>
" standard C "escape sequences" (such as \n for newline or \" for a quote).<br>
<br>
So actually I guess we should support a few more things which are in the C<br>
language but not part of the RIB syntax:<br>
<br>
\a<br>
\v<br>
\'<br>
\?<br>
\ooo<br>
\x hh<br>
\x hhh<br>
(According to <a href="http://msdn.microsoft.com/en-us/library/h21280bw%28VS.80%29.aspx" target="_blank" rel="nofollow">http://msdn.microsoft.com/en-us/library/h21280bw(VS.80).aspx</a>)<br></blockquote><div><br>I agree with. this. I misinterpreted the specs. Anyway most of this new escape characteres are easy to add.<font face="Arial"><b><span style="font-size: 14px;"><span id="IDA2MXYE" style=""></span></span></b></font></div>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
I've also got some suggestions to make the code a bit easier to read. To begin<br>
with, please consider putting this chunk of code into its own function<br>
(eg, LoadString() or whatever). CqShaderVM::LoadProgram() is a monster<br>
function already, let's not make it any huger!<br>
<br></blockquote><div>Yes, I'll do it. <br></div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
Other suggestions follow:<br>
<br>
> diff --git a/libs/shadervm/shadervm.cpp b/libs/shadervm/shadervm.cpp<br>
> index 225670a..a8d693e 100644<br>
> --- a/libs/shadervm/shadervm.cpp<br>
> +++ b/libs/shadervm/shadervm.cpp<br>
> @@ -1279,11 +1279,89 @@ void CqShaderVM::LoadProgram( std::istream* pFile )<br>
> ( *pFile ) >> std::ws;<br>
> char c;<br>
> CqString s( "" );<br>
> + bool escapeChar = false, escapeNumeric = false;<br>
> + CqString capturedNumber( "" );<br>
> +<br>
> pFile->get<br>
> ();<br>
> - while ( ( c = pFile->get<br>
> - () ) != '"' )<br>
> - s += c;<br>
> + while ( (( c = pFile->get() ) != '"') || escapeChar )<br>
> + {<br>
> +<br>
> + if (escapeNumeric)<br>
> + {<br>
> + if(c < '0' or c > '9')<br>
> + {<br>
> + char number = strtoul(capturedNumber.c_str(), NULL, 8);<br>
> + s+= number;<br>
> + escapeChar = false;<br>
> + escapeNumeric = false;<br>
> + capturedNumber = "";<br>
> + }<br>
> + }<br>
> + if(escapeChar)<br>
> + {<br>
> + //Treatment for escape char<br>
> + switch(c){<br>
<br>
Please use opening braces on their own line for consistency with the rest of<br>
the codebase.<br></blockquote><div><br>I installed Artistic style to check the code style but.. Where is the style file?</div><blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
> + case 'n':<br>
> + s += '\n';<br>
> + escapeChar = false;<br>
> + break;<br>
> + case 'r':<br>
> + s += '\r';<br>
> + escapeChar = false;<br>
> + break;<br>
> + case 't':<br>
> + s += '\t';<br>
> + escapeChar = false;<br>
> + break;<br>
> + case 'b':<br>
> + s += '\b';<br>
> + escapeChar = false;<br>
> + break;<br>
> + case 'f':<br>
> + s += '\f';<br>
> + escapeChar = false;<br>
> + break;<br>
> + case '"':<br>
> + s += '"';<br>
> + escapeChar = false;<br>
> + break;<br>
> + case '\\':<br>
> + s += '\\';<br>
> + escapeChar = false;<br>
> + break;<br>
> + case '0':<br>
> + case '1':<br>
> + case '2':<br>
> + case '3':<br>
> + case '4':<br>
> + case '5':<br>
> + case '6':<br>
> + case '7':<br>
> + case '8':<br>
> + case '9':<br>
> +<br>
> + capturedNumber += c;<br>
> + if (capturedNumber.length() == 3)<br>
> + {<br>
> + escapeNumeric = false;<br>
> + escapeChar = false;<br>
> + s += (char)strtoul(capturedNumber.c_str(), NULL, 8);<br>
> + }<br>
> + else<br>
> + escapeNumeric = true;<br>
> + break;<br>
<br>
When I was reading through this code, I got rather confused by the handling of<br>
capturedNumber, since it appears in two different places and there's two<br>
different places where it's converted to a char using strtoul(). I suggest<br>
just using a nested loop inside this part of the case statement to confine<br>
octal handling logic to a single place in the code. When you encounter the<br>
end of the octal number you can use the function pFile->unget() to put back<br>
the terminating character into the input stream. (For an example of something<br>
very similar, see CqRibLexer::readString() in riblexer.cpp.)<br></blockquote><div><br>I wrote the code in this way to avoid using file->unget() when the length of the number is less than 3. Anyway I´ll check it<br></div>
<blockquote class="gmail_quote" style="border-left: 1px solid rgb(204, 204, 204); margin: 0pt 0pt 0pt 0.8ex; padding-left: 1ex;">
<br>
> + default :<br>
> + escapeChar = false;<br>
> + break;<br>
> + }<br>
> + }<br>
> + else<br>
> + if (c == '\\')<br>
> + escapeChar = true;<br>
> + else<br>
> + s += c;<br>
> + }<br>
> AddString( s.c_str(), pProgramArea );<br>
> }<br>
> break;<br>
<br>
Just attach any changes as a new patch, thanks :-)<br>
~Chris.<br>
<br>
------------------------------------------------------------------------------<br>
Come build with us! The BlackBerry&reg; Developer Conference in SF, CA<br>
is the only developer event you need to attend this year. Jumpstart your<br>
developing skills, take BlackBerry mobile applications to market and stay<br>
ahead of the curve. Join us from November 9&#45;12, 2009. Register now&#33;<br>
<a href="http://p.sf.net/sfu/devconf" target="_blank" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>
_______________________________________________<br>
Aqsis-development mailing list<br>
<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25694651&i=1" target="_top" rel="nofollow">Aqsis-development@...</a><br>
<a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_blank" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br>
</blockquote></div><br>
<br />------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br />_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25694651&i=2" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25689806Re: patch for bug id 2813534 "Escaped strings in RSL not working"2009-09-30T14:58:17Z2009-09-30T14:58:17ZChris Foster-5Hi Ricardo,
<br><br>On Sun, Sep 27, 2009 at 04:12:57PM +0100, Ricardo Mayor wrote:
<div class='shrinkable-quote'><br>> Hello,
<br>> Find attached a patch for the bug "*Escaped strings in RSL not working".
<br>> According *to ri specs, the supported escape string are
<br>>
<br>> \n linefeed (newline)
<br>> \r carriage return
<br>> \t horizontal tab
<br>> \b backspace
<br>> \f form feed
<br>> \\ backslash
<br>> \" double quote
<br>> \ddd character code ddd (octal)
<br>> \newline no character ? both are ignored
</div><br>I think these may be the escape sequences specified in the part of the RISpec
<br>which deals with the RIB format, not the renderman shading language?
<br><br>Here's what I see in my copy of the RISpec (around page 117):
<br><br>" Strings are used to name external objects (texture maps, for
<br>example). String literals (or
<br>" constants) are enclosed in double quotes, as in the C language, and
<br>may contain any of the
<br>" standard C "escape sequences" (such as \n for newline or \" for a quote).
<br><br>So actually I guess we should support a few more things which are in the C
<br>language but not part of the RIB syntax:
<br><br>\a
<br>\v
<br>\'
<br>\?
<br>\ooo
<br>\x hh
<br>\x hhh
<br>(According to <a href="http://msdn.microsoft.com/en-us/library/h21280bw(VS.80).aspx" target="_top" rel="nofollow">http://msdn.microsoft.com/en-us/library/h21280bw(VS.80).aspx</a>)
<br><br>I've also got some suggestions to make the code a bit easier to read. To begin
<br>with, please consider putting this chunk of code into its own function
<br>(eg, LoadString() or whatever). CqShaderVM::LoadProgram() is a monster
<br>function already, let's not make it any huger!
<br><br>Other suggestions follow:
<br><div class='shrinkable-quote'><br>> diff --git a/libs/shadervm/shadervm.cpp b/libs/shadervm/shadervm.cpp
<br>> index 225670a..a8d693e 100644
<br>> --- a/libs/shadervm/shadervm.cpp
<br>> +++ b/libs/shadervm/shadervm.cpp
<br>> @@ -1279,11 +1279,89 @@ void CqShaderVM::LoadProgram( std::istream* pFile )
<br>> ( *pFile ) >> std::ws;
<br>> char c;
<br>> CqString s( "" );
<br>> + bool escapeChar = false, escapeNumeric = false;
<br>> + CqString capturedNumber( "" );
<br>> +
<br>> pFile->get
<br>> ();
<br>> - while ( ( c = pFile->get
<br>> - () ) != '"' )
<br>> - s += c;
<br>> + while ( (( c = pFile->get() ) != '"') || escapeChar )
<br>> + {
<br>> +
<br>> + if (escapeNumeric)
<br>> + {
<br>> + if(c < '0' or c > '9')
<br>> + {
<br>> + char number = strtoul(capturedNumber.c_str(), NULL, 8);
<br>> + s+= number;
<br>> + escapeChar = false;
<br>> + escapeNumeric = false;
<br>> + capturedNumber = "";
<br>> + }
<br>> + }
<br>> + if(escapeChar)
<br>> + {
<br>> + //Treatment for escape char
<br>> + switch(c){
</div><br>Please use opening braces on their own line for consistency with the rest of
<br>the codebase.
<br><div class='shrinkable-quote'><br>> + case 'n':
<br>> + s += '\n';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case 'r':
<br>> + s += '\r';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case 't':
<br>> + s += '\t';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case 'b':
<br>> + s += '\b';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case 'f':
<br>> + s += '\f';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case '"':
<br>> + s += '"';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case '\\':
<br>> + s += '\\';
<br>> + escapeChar = false;
<br>> + break;
<br>> + case '0':
<br>> + case '1':
<br>> + case '2':
<br>> + case '3':
<br>> + case '4':
<br>> + case '5':
<br>> + case '6':
<br>> + case '7':
<br>> + case '8':
<br>> + case '9':
<br>> +
<br>> + capturedNumber += c;
<br>> + if (capturedNumber.length() == 3)
<br>> + {
<br>> + escapeNumeric = false;
<br>> + escapeChar = false;
<br>> + s += (char)strtoul(capturedNumber.c_str(), NULL, 8);
<br>> + }
<br>> + else
<br>> + escapeNumeric = true;
<br>> + break;
</div><br>When I was reading through this code, I got rather confused by the handling of
<br>capturedNumber, since it appears in two different places and there's two
<br>different places where it's converted to a char using strtoul(). I suggest
<br>just using a nested loop inside this part of the case statement to confine
<br>octal handling logic to a single place in the code. When you encounter the
<br>end of the octal number you can use the function pFile->unget() to put back
<br>the terminating character into the input stream. (For an example of something
<br>very similar, see CqRibLexer::readString() in riblexer.cpp.)
<br><div class='shrinkable-quote'><br>> + default :
<br>> + escapeChar = false;
<br>> + break;
<br>> + }
<br>> + }
<br>> + else
<br>> + if (c == '\\')
<br>> + escapeChar = true;
<br>> + else
<br>> + s += c;
<br>> + }
<br>> AddString( s.c_str(), pProgramArea );
<br>> }
<br>> break;
</div><br>Just attach any changes as a new patch, thanks :-)
<br>~Chris.
<br><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25689806&i=0" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25638453Re: constant vs smooth shading interpolation2009-09-27T15:51:08Z2009-09-27T15:51:08ZChris Foster-5On Mon, Sep 28, 2009 at 7:23 AM, Matthias Baas <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25638453&i=0" target="_top" rel="nofollow">matthias.baas@...</a>> wrote:
<br>> True, in particular when it's the default now. I suppose improving
<br>> constant interpolation only makes sense when there is a noticeable
<br>> performance benefit. I just did a quick test with that scene from the
<br>> bug report and rendered it using a very low shading rate (to create more
<br>> micro polys). Smooth shading was only slightly slower and I suppose that
<br>> difference will get even smaller once there is more complex geometry and
<br>> shading involved. So, just using smooth interpolation is fine by me. :)
<br><br>Note that the smooth shading implementation has changed completely
<br>since 1.4, but I think the conclusion that the speed hit is fairly
<br>minor (when compared to other costs of complex scenes) tends to be
<br>generally true. Overall I think that defaulting to smooth is a win
<br>compared to dealing with the confusing artefacts which come up every
<br>now and then with constant interpolation.
<br><br>Note also that the hit depth always needs better than constant
<br>interpolation for proper hiding (in aqsis-1.4 it was a kind of halfway
<br>smooth interpolation). In the new implementation smooth interpolation
<br>is used for the depth so some of the cost of smooth interpolation is
<br>always present due to the need to calculate z accurately (that is, the
<br>cost of reverse bilinear lookup from the hit location to micropolygon
<br>u,v coordinates).
<br><br>~Chris.
<br><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25638453&i=1" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25637751Re: constant vs smooth shading interpolation2009-09-27T14:23:00Z2009-09-27T14:23:00ZMatthias Baas-3Hi Chris,
<br><br>Chris Foster wrote:
<br>> I'm sure we could come up with a number of complex workarounds, but
<br>> TBH I don't know why we'd bother because we already have one: it's
<br>> called smooth shading interpolation!
<br><br>True, in particular when it's the default now. I suppose improving
<br>constant interpolation only makes sense when there is a noticeable
<br>performance benefit. I just did a quick test with that scene from the
<br>bug report and rendered it using a very low shading rate (to create more
<br>micro polys). Smooth shading was only slightly slower and I suppose that
<br>difference will get even smaller once there is more complex geometry and
<br>shading involved. So, just using smooth interpolation is fine by me. :)
<br><br>Cheers,
<br><br>- Matthias -
<br><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25637751&i=0" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25634434patch for bug id 2813534 "Escaped strings in RSL not working"2009-09-27T08:12:57Z2009-09-27T08:12:57ZRicardo MayorHello,<br>Find attached a patch for the bug "<span><strong>Escaped strings in RSL not working". <br><span style="font-weight: normal;">According </span></strong></span>to ri specs, the supported escape string are<br>
<br>\n linefeed (newline)<br>\r carriage return<br>\t horizontal tab<br>\b backspace<br>\f form feed<br>\\ backslash<br>\” double quote<br>\ddd character code ddd (octal)<br>\newline no character – both are ignored<br>
<br>Only the last one was currently supported. The attached patch adds support for all the cases.<br><br>Best regards<br>
<br /><tt>[escape_char.patch]</tt><br /><hr align="left" width="300" /><tt>diff --git a/libs/shadervm/shadervm.cpp b/libs/shadervm/shadervm.cpp
<br>index 225670a..a8d693e 100644
<br>--- a/libs/shadervm/shadervm.cpp
<br>+++ b/libs/shadervm/shadervm.cpp
<br>@@ -1279,11 +1279,89 @@ void CqShaderVM::LoadProgram( std::istream* pFile )
<br> ( *pFile ) >> std::ws;
<br> char c;
<br> CqString s( "" );
<br>+ bool escapeChar = false, escapeNumeric = false;
<br>+ CqString capturedNumber( "" );
<br>+
<br> pFile->get
<br> ();
<br>- while ( ( c = pFile->get
<br>- () ) != '"' )
<br>- s += c;
<br>+ while ( (( c = pFile->get() ) != '"') || escapeChar )
<br>+ {
<br>+
<br>+ if (escapeNumeric)
<br>+ {
<br>+ if(c < '0' or c > '9')
<br>+ {
<br>+ char number = strtoul(capturedNumber.c_str(), NULL, 8);
<br>+ s+= number;
<br>+ escapeChar = false;
<br>+ escapeNumeric = false;
<br>+ capturedNumber = "";
<br>+ }
<br>+ }
<br>+ if(escapeChar)
<br>+ {
<br>+ //Treatment for escape char
<br>+ switch(c){
<br>+ case 'n':
<br>+ s += '\n';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case 'r':
<br>+ s += '\r';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case 't':
<br>+ s += '\t';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case 'b':
<br>+ s += '\b';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case 'f':
<br>+ s += '\f';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case '"':
<br>+ s += '"';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case '\\':
<br>+ s += '\\';
<br>+ escapeChar = false;
<br>+ break;
<br>+ case '0':
<br>+ case '1':
<br>+ case '2':
<br>+ case '3':
<br>+ case '4':
<br>+ case '5':
<br>+ case '6':
<br>+ case '7':
<br>+ case '8':
<br>+ case '9':
<br>+
<br>+ capturedNumber += c;
<br>+ if (capturedNumber.length() == 3)
<br>+ {
<br>+ escapeNumeric = false;
<br>+ escapeChar = false;
<br>+ s += (char)strtoul(capturedNumber.c_str(), NULL, 8);
<br>+ }
<br>+ else
<br>+ escapeNumeric = true;
<br>+ break;
<br>+ default :
<br>+ escapeChar = false;
<br>+ break;
<br>+ }
<br>+ }
<br>+ else
<br>+ if (c == '\\')
<br>+ escapeChar = true;
<br>+ else
<br>+ s += c;
<br>+ }
<br> AddString( s.c_str(), pProgramArea );
<br> }
<br> break;
<br></tt><hr align="left" width="300" /><br />------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br />_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25634434&i=0" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25614459Re: Still no luck running Aqsis 1.5.0 on Tiger2009-09-25T08:48:53Z2009-09-25T08:48:53ZTobias Sauerwein-2Great, its a little dated, but good to know that it works.
<br><br>On Thu, Sep 24, 2009 at 10:35 PM, Matthias Baas <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25614459&i=0" target="_top" rel="nofollow">matthias.baas@...</a>> wrote:
<div class='shrinkable-quote'><br>> Hi Tobias,
<br>>
<br>>> I'm not sure if our nightlies are propper universal builds.
<br>>> Give this one a try, please ( <a href="http://cgtobi.de/aqsis/Aqsis-1.5.0-Darwin.dmg" target="_top" rel="nofollow">http://cgtobi.de/aqsis/Aqsis-1.5.0-Darwin.dmg</a> )
<br>>
<br>> ok, that one works! Thanks! :)
<br>>
<br>> - Matthias -
<br>>
<br>>
<br>> ------------------------------------------------------------------------------
<br>> Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>> is the only developer event you need to attend this year. Jumpstart your
<br>> developing skills, take BlackBerry mobile applications to market and stay
<br>> ahead of the curve. Join us from November 9-12, 2009. Register now!
<br>> <a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>> _______________________________________________
<br>> Aqsis-development mailing list
<br>> <a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25614459&i=1" target="_top" rel="nofollow">Aqsis-development@...</a>
<br>> <a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br>>
</div><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25614459&i=2" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25603026Re: constant vs smooth shading interpolation2009-09-24T14:29:16Z2009-09-24T14:29:16ZChris Foster-5On Fri, Sep 25, 2009 at 7:02 AM, Matthias Baas <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25603026&i=0" target="_top" rel="nofollow">matthias.baas@...</a>> wrote:
<div class='shrinkable-quote'><br>> Hi Chris,
<br>>
<br>>> Yes, that is basically the problem. It's important to realise that the value
<br>>> of ShadingInterpolation in no way influences how the shading itself is done,
<br>>> which is _always_ at the vertices of a micropolygon grid. Instead, the
<br>>> shading interpolation effects how the results of running the shader program
<br>>> are interpolated across the faces of individual micropolygons:
<br>>>
<br>>> ShadingInterpolation "constant":
<br>>> Constant interpolation means that each micropolygon is flat shaded, with the
<br>>> colour coming from a single vertex. The micropolygon at position (iu,iv) on
<br>>> a grid is shaded with the colour from vertex (iu,iv) so that there's
<br>>> effectively half a micropolygon width offset between the centre of the
<br>>> micropolygon and the position at which the shading result was computed.
<br>>
<br>> ok, so this means there is always a row and column of vertices that are
<br>> unused and I guess from the renderer's point of view it would be just as
<br>> correct if micropoly (iu,iv) would get the color from vertex (iu+1,iv),
<br>> (iu,iv+1) or (iu+1, iv+1) (i.e. from any of the other micropoly vertices).
<br>> Couldn't this be exploited to make the shading across grids more
<br>> uniform?
</div><br>Yes, I could imagine doing this to some extent but it will never work
<br>in full generality. (To see why, consider trying to shade a
<br>nonorientable manifold like the Mobius strip!) In practise it would
<br>be sufficient that the places where the directions of increasing u and
<br>v are inconsistent are where the surface is tangent to the view
<br>direction.
<br><br><div class='shrinkable-quote'><br>> Maybe the orientation of the grid in space could be used to
<br>> decide whether the "lower" or the "upper" row/column of vertices is
<br>> ignored. For example, if the vertex coordinates in one direction (say x
<br>> direction) are increasing, then use iu, otherwise use iu+1 and the same
<br>> in v direction. What I mean is, the micropolys from a grid that is
<br>> rotated around by 180 degrees should then get the same shading values
<br>> than the micropolys from the unrotated grid and that artifact in the bug
<br>> report would hopefully go away, even in constant shading interpolation mode.
<br>> Or alternatively, would it be an option to shader the micropolys at
<br>> their centers?
</div><br>It's possible in principle, but probably not desirable in practise: it
<br>requires that surface shading run on a different grid from the
<br>displacement shader, which at the least implies a cost in code
<br>complexity.
<br><br>I'm sure we could come up with a number of complex workarounds, but
<br>TBH I don't know why we'd bother because we already have one: it's
<br>called smooth shading interpolation!
<br><br>~Chris.
<br><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25603026&i=1" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25602635Re: constant vs smooth shading interpolation2009-09-24T14:02:47Z2009-09-24T14:02:47ZMatthias Baas-3Hi Chris,
<br><div class='shrinkable-quote'><br>> Yes, that is basically the problem. It's important to realise that the value
<br>> of ShadingInterpolation in no way influences how the shading itself is done,
<br>> which is _always_ at the vertices of a micropolygon grid. Instead, the
<br>> shading interpolation effects how the results of running the shader program
<br>> are interpolated across the faces of individual micropolygons:
<br>>
<br>> ShadingInterpolation "constant":
<br>> Constant interpolation means that each micropolygon is flat shaded, with the
<br>> colour coming from a single vertex. The micropolygon at position (iu,iv) on
<br>> a grid is shaded with the colour from vertex (iu,iv) so that there's
<br>> effectively half a micropolygon width offset between the centre of the
<br>> micropolygon and the position at which the shading result was computed.
</div><br>ok, so this means there is always a row and column of vertices that are
<br>unused and I guess from the renderer's point of view it would be just as
<br>correct if micropoly (iu,iv) would get the color from vertex (iu+1,iv),
<br>(iu,iv+1) or (iu+1, iv+1) (i.e. from any of the other micropoly vertices).
<br>Couldn't this be exploited to make the shading across grids more
<br>uniform? Maybe the orientation of the grid in space could be used to
<br>decide whether the "lower" or the "upper" row/column of vertices is
<br>ignored. For example, if the vertex coordinates in one direction (say x
<br>direction) are increasing, then use iu, otherwise use iu+1 and the same
<br>in v direction. What I mean is, the micropolys from a grid that is
<br>rotated around by 180 degrees should then get the same shading values
<br>than the micropolys from the unrotated grid and that artifact in the bug
<br>report would hopefully go away, even in constant shading interpolation mode.
<br>Or alternatively, would it be an option to shader the micropolys at
<br>their centers?
<br><div class='shrinkable-quote'><br>>> that triangle almost looks as if it was smooth shaded. It seems Aqsis
<br>>> produces far more micro polygons than necessary (according to the stats
<br>>> it's 1233 vs 902). How does it determine the number of micro polygons to
<br>>> create?
<br>>
<br>> Aqsis computes the length of the primitive in the u-direction, and the length
<br>> in the v-direction in raster space. It then divides these by
<br>> sqrt(ShadingRate) to estimate the number of micropolygons needed in each
<br>> direction. This is appropriate and accurate when the sides of a grid are
<br>> nearly perpendicular.
<br>>
<br>> It's probable that this goes wrong in your test scene because the triangle at
<br>> the back is quite long and skinny.
</div><br>This is already considered to be long and skinny? Then I suppose most
<br>triangles are. ;)
<br>Maybe I can eventually do some more test scenes to investigate this further.
<br><br>Cheers,
<br><br>- Matthias -
<br><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25602635&i=0" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25602204Re: Still no luck running Aqsis 1.5.0 on Tiger2009-09-24T13:35:58Z2009-09-24T13:35:58ZMatthias Baas-3Hi Tobias,
<br><br>> I'm not sure if our nightlies are propper universal builds.
<br>> Give this one a try, please ( <a href="http://cgtobi.de/aqsis/Aqsis-1.5.0-Darwin.dmg" target="_top" rel="nofollow">http://cgtobi.de/aqsis/Aqsis-1.5.0-Darwin.dmg</a> )
<br><br>ok, that one works! Thanks! :)
<br><br>- Matthias -
<br><br><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25602204&i=0" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25569729Re: [Aqsis-commits] [SCM] Aqsis Renderer branch, master, updated. CVS_Tip-2554-gef5ab6d2009-09-22T17:44:10Z2009-09-22T17:44:10ZChris Foster-5On Mon, Sep 21, 2009 at 7:16 PM, Paul Gregory
<br><<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25569729&i=0" target="_top" rel="nofollow">pgregory@...</a>> wrote:
<br>> - Log -----------------------------------------------------------------
<br>> commit ef5ab6d1443e7284e096c0803abecde5aff9e1df
<br>> Author: Paul Gregory <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25569729&i=1" target="_top" rel="nofollow">pgregory@...</a>>
<br>> Date: Mon Sep 21 09:26:54 2009 +0100
<br>>
<br>> Fix bug #2738390.
<br>[...]
<br><br>This looks nice Paul; it's far cleaner and simpler than before :-)
<br><br>One question:
<br><div class='shrinkable-quote'><br>> diff --git a/libs/core/api/ri.cpp b/libs/core/api/ri.cpp
<br>> index 4c7e958..4874b3e 100644
<br>> --- a/libs/core/api/ri.cpp
<br>> +++ b/libs/core/api/ri.cpp
<br>> @@ -1085,6 +1085,7 @@ RtVoid RiProjectionV( RtToken name, PARAMETERLIST )
<br>> }
<br>> }
<br>> // TODO: need to get the current transformation so that it can be added to the screen transformation.
<br>> + QGetRenderContext() ->SetpreProjectionTransform( QGetRenderContext() ->ptransCurrent() );
<br>> QGetRenderContext() ->ptransSetTime( CqMatrix() );
</div><br>Is this TODO now done? Did it refer to the pre-projection transformation?
<br><br>~Chris.
<br><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25569729&i=2" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25569667Re: constant vs smooth shading interpolation2009-09-22T17:35:26Z2009-09-22T17:35:26ZChris Foster-5On Wed, Sep 23, 2009 at 9:29 AM, Chris Foster <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25569667&i=0" target="_top" rel="nofollow">chris42f@...</a>> wrote:
<br>> I think this would be fairly easy to fix (at least for bilinear patches) using
<br>> the magnitude of the cross product of the two sides. This has an appropriate
<br>> scaling factor of sin(angle_between_the_sides) so that it computes a much
<br>> better estimate of the expected area of the bilinear patch.
<br><br>Hmm, or maybe this isn't really the right thing to do. It would result in a
<br>more accurate micropolygon *area*, but the tradeoff would be micropolygons
<br>which are relatively large in linear dimension with possible artifacts...
<br><br><br>One thing which I'd guess is definitely worth doing is to improve the way that
<br>triangles are turned into patches. Currently, a triangle is turned into a
<br>patch with a random orientation, but this can result in unnecessarily long
<br>skinny patches.
<br><br>Consider the triangle
<br><br> x
<br> |\
<br> | \
<br> | \
<br> x---x
<br><br>one way to turn this into a patch with a phantom fourth vertex (indicated by
<br>'o') is
<br><br> x
<br> |\
<br> | \
<br> | \
<br> x---x
<br> \ |
<br> \ |
<br> \|
<br> o
<br><br>but this results in long skinny micropolygons (micropolys have the same shape
<br>as the parent patch). Alternatively, the original triangle could be turned
<br>into the following:
<br><br> x---o
<br> |\ |
<br> | \ |
<br> | \|
<br> x---x
<br><br>which has square micropolygons which are much better.
<br><br><br>Looking at the current code, I see that this has actually been implemented in
<br>the past, but was commented out in 2006, apparently for the reason:
<br><br>/* Comment this out for now as it breaks Bug948827 in a strange way,
<br>needs more investigation */
<br><br>presumably this wouldn't be too hard to fix.
<br><br>~Chris.
<br><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25569667&i=1" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>tag:old.nabble.com,2006:post-25569152Re: constant vs smooth shading interpolation2009-09-22T16:29:42Z2009-09-22T16:29:42ZChris Foster-5Hi Matthias,
<br><br>On Wed, Sep 23, 2009 at 7:44 AM, Matthias Baas <<a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25569152&i=0" target="_top" rel="nofollow">matthias.baas@...</a>> wrote:
<div class='shrinkable-quote'><br>> Hi,
<br>>
<br>> this is basically a followup to my bug report 2862585
<br>> (<a href="http://sourceforge.net/tracker/?func=detail&aid=2862585&group_id=25264&atid=383970" target="_top" rel="nofollow">http://sourceforge.net/tracker/?func=detail&aid=2862585&group_id=25264&atid=383970</a>).
<br>> I thought I rather post here instead of adding a reply to the bug report
<br>> (which is closed anyway). I think Chris' solution to switch to smooth
<br>> shading interpolation is fair enough, I'm just posting to make sure I
<br>> understand the problem correctly.
<br>>
<br>> Basically, there's one question left for me: in constant interpolation
<br>> mode, at what point is the shading actually done? Is it just one of the
<br>> micro polygon vertex? If so, which one? Is that the problem in my
<br>> example scene that where the two triangles meet, the micro polygons are
<br>> oriented in the "opposite" way which means the shading locations are two
<br>> micro polygons apart instead of just one which leads to this discontinuity?
</div><br>Yes, that is basically the problem. It's important to realise that the value
<br>of ShadingInterpolation in no way influences how the shading itself is done,
<br>which is _always_ at the vertices of a micropolygon grid. Instead, the
<br>shading interpolation effects how the results of running the shader program
<br>are interpolated across the faces of individual micropolygons:
<br><br>ShadingInterpolation "constant":
<br> Constant interpolation means that each micropolygon is flat shaded, with the
<br> colour coming from a single vertex. The micropolygon at position (iu,iv) on
<br> a grid is shaded with the colour from vertex (iu,iv) so that there's
<br> effectively half a micropolygon width offset between the centre of the
<br> micropolygon and the position at which the shading result was computed.
<br><br>ShadingInterpolation "smooth":
<br> Smooth (or gouraud) shading causes the colour of the four vertices to be
<br> interpolated bilinearly across the face of the micropolygon. In this case
<br> there's no offset bias between the position of the shading point and the
<br> effects it has in the output image.
<br><br>Since a grid has no information about its neighbours, it can happen that the
<br>grid indices iu and iv could increase in different directions. As you
<br>noticed, this is a problem for constant interpolation because it causes a
<br>micropolygon's width of "shear" in the output image.
<br><br>> I also rendered that scene using PRMan and 3Delight. Both of them
<br>> produce micro polygons whose sides are parallel to the initial polygon
<br>> (which means there is no artifact in this example). Does that mean they
<br>> don't split up the quad into two triangles but rather handle quads
<br>> separately?
<br><br>Most likely I would say. This would just require detecting when a general
<br>polygon was actually a convex quad, which is pretty easy I think.
<br><div class='shrinkable-quote'><br>> When I further played around with that scene I found another thing that
<br>> looks odd to me. When you replace the polygon by this one (which just
<br>> has an additional vertex):
<br>>
<br>> PointsGeneralPolygons [1] [5] [0 1 3 2 4] "P" [-1 0 1 1 0 1 -1 0 -1 1
<br>> 0 -1 -0.5 0 0]
<br>>
<br>> and increase the shading rate to 80 (so that you can easily see what's
<br>> going on), then the results of Aqsis and 3Delight in the lower part of
<br>> the image actually look similar (the micro polygons are oriented
<br>> differently but after all, that's up to the renderer). The odd thing is
<br>> the triangle on the far side. In 3Delight's case, you can easily spot
<br>> the micro polygons just like in the lower part. They even seem to be a
<br>> bit larger which I guess is just because they are further away (and
<br>> maybe just appear larger? I didn't count pixels) but in Aqsis's case,
<br>> that triangle almost looks as if it was smooth shaded. It seems Aqsis
<br>> produces far more micro polygons than necessary (according to the stats
<br>> it's 1233 vs 902). How does it determine the number of micro polygons to
<br>> create?
</div><br>Aqsis computes the length of the primitive in the u-direction, and the length
<br>in the v-direction in raster space. It then divides these by
<br>sqrt(ShadingRate) to estimate the number of micropolygons needed in each
<br>direction. This is appropriate and accurate when the sides of a grid are
<br>nearly perpendicular.
<br><br>It's probable that this goes wrong in your test scene because the triangle at
<br>the back is quite long and skinny. The area of the micropolygons arising from
<br>such a triangle is much smaller than what the simple estimate based on
<br>perpendicular u and v directions would suggest. Therefore aqsis overestimates
<br>the dicing rate.
<br><br>I think this would be fairly easy to fix (at least for bilinear patches) using
<br>the magnitude of the cross product of the two sides. This has an appropriate
<br>scaling factor of sin(angle_between_the_sides) so that it computes a much
<br>better estimate of the expected area of the bilinear patch.
<br><br>Cheers,
<br>~Chris.
<br><br>------------------------------------------------------------------------------
<br>Come build with us! The BlackBerry® Developer Conference in SF, CA
<br>is the only developer event you need to attend this year. Jumpstart your
<br>developing skills, take BlackBerry mobile applications to market and stay
<br>ahead of the curve. Join us from November 9-12, 2009. Register now!
<br><a href="http://p.sf.net/sfu/devconf" target="_top" rel="nofollow">http://p.sf.net/sfu/devconf</a><br>_______________________________________________
<br>Aqsis-development mailing list
<br><a href="http://old.nabble.com/user/SendEmail.jtp?type=post&post=25569152&i=1" target="_top" rel="nofollow">Aqsis-development@...</a>
<br><a href="https://lists.sourceforge.net/lists/listinfo/aqsis-development" target="_top" rel="nofollow">https://lists.sourceforge.net/lists/listinfo/aqsis-development</a><br><p>From forum: <a href="http://old.nabble.com/aqsis-development-f3702.html" embed="fixTarget[3702]" target="_top" >aqsis-development</a></p>