Broke again - this time in jsgc.cpp

[Rich Walsh]

Once again they forgot we exist, so there was no OS/2 memory allocation
code in the new-and-improved Javascript garbage collector.  I filed
Bug 595326 and attached a patch to fix the omission.

BTW... the collossal instability I reported for last night's build
has miraculously disappeared in tonight's build.


--
== == almost usable email address:  Rich AT E-vertise.Com == ==
___________________________________________________________________
                |
                |         DragText v3.9 with NLS support
Rich Walsh      |   A Distinctly Different Desktop Enhancement
Ft Myers, FL    |         http://e-vertise.com/dragtext/
___________________________________________________________________

_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Steve Wendt]

On 10/29/09 09:34 pm, Rich Walsh wrote:

> Once again they forgot we exist, so there was no OS/2 memory allocation
> code in the new-and-improved Javascript garbage collector.  I filed
> Bug 595326 and attached a patch to fix the omission.

Dyslexia strikes us all from time to time...
https://bugzilla.mozilla.org/show_bug.cgi?id=525396
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Carl Gehr-7]

On 10/30/09 12:34 am, Rich Walsh wrote:
> Once again they forgot we exist, so there was no OS/2 memory allocation
> code in the new-and-improved Javascript garbage collector.  I filed
> Bug 595326 and attached a patch to fix the omission.
>
> BTW... the collossal instability I reported for last night's build
> has miraculously disappeared in tonight's build.

Sorry if I'm out of it, but 'what' is 'broke' again?
[I don't want to use it if it is broken.  But if it's not something I
use, I don't have to worry either...
--
TIA,
Carl
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Dave Yeo-3]

On 10/30/09 07:22 am, Carl Gehr wrote:
> On 10/30/09 12:34 am, Rich Walsh wrote:
>> Once again they forgot we exist, so there was no OS/2 memory allocation
>> code in the new-and-improved Javascript garbage collector. I filed
>> Bug 595326 and attached a patch to fix the omission.
>>
>> BTW... the collossal instability I reported for last night's build
>> has miraculously disappeared in tonight's build.
>
> Sorry if I'm out of it, but 'what' is 'broke' again?

The actual building due to new code using functions that are not
available on OS/2

> [I don't want to use it if it is broken. But if it's not something I
> use, I don't have to worry either...

I wouldn't worry about this and also Rich's bad build. Occasionally the
cutting edge development code gets unstable, usually this is fixed
pretty quick.
Dave
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Carl Gehr-7]

On 10/30/09 11:25 am, Dave Yeo wrote:

> On 10/30/09 07:22 am, Carl Gehr wrote:
>> On 10/30/09 12:34 am, Rich Walsh wrote:
>>> Once again they forgot we exist, so there was no OS/2 memory allocation
>>> code in the new-and-improved Javascript garbage collector. I filed
>>> Bug 595326 and attached a patch to fix the omission.
>>>
>>> BTW... the collossal instability I reported for last night's build
>>> has miraculously disappeared in tonight's build.
>>
>> Sorry if I'm out of it, but 'what' is 'broke' again?
>
> The actual building due to new code using functions that are not
> available on OS/2
>

No, I guess I was not clear...
Building 'What'...?  SeaMonkey?  FireFox?  Javascript that impacts all
Mozilla? Something else?

No where that I can see tells _me_ what 'it' is?
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Steve Wendt]

On 10/30/2009 12:36 PM, Carl Gehr wrote:

>>>> Once again they forgot we exist, so there was no OS/2 memory allocation
>>>> code in the new-and-improved Javascript garbage collector. I filed
>>>> Bug 595326 and attached a patch to fix the omission.
>>>
>>> Sorry if I'm out of it, but 'what' is 'broke' again?
>
> Building 'What'...? SeaMonkey? FireFox? Javascript that impacts all
> Mozilla? Something else?

It's in the code that impacts *future* versions of all Mozilla products,
and has nothing at all to do with current versions like Seamonkey 2.0 or
Firefox 3.5.  This newsgroup is really a development forum (hence the
"dev" in the name), and this is not an end-user friendly topic.  Feel
free to (and please do) ignore such threads.  ;-)
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Peter Weilbacher]

On 30/10/09 05:34, Rich Walsh wrote:
> Once again they forgot we exist, so there was no OS/2 memory allocation
> code in the new-and-improved Javascript garbage collector.  I filed
> Bug 595326 and attached a patch to fix the omission.

I didn't realize we were ever unbroken. What's the trick with libffi?
The info in bug 522118 doesn't really help. (That's probably only the
first problem I found while trying to produce FF 3.6b1.)
--
Please     | Official Warpzilla Ports: http://www.mozilla.org/ports/os2/
reply in   |
newsgroup  |          Enhanced OS/2 builds: http://pmw-warpzilla.sf.net/
     Steve's Warpzilla Tips: http://www.os2bbs.com/os2news/Warpzilla.html
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Dave Yeo-3]

On 10/30/09 03:25 pm, Peter Weilbacher wrote:
> On 30/10/09 05:34, Rich Walsh wrote:
>> Once again they forgot we exist, so there was no OS/2 memory allocation
>> code in the new-and-improved Javascript garbage collector.  I filed
>> Bug 595326 and attached a patch to fix the omission.
>
> I didn't realize we were ever unbroken. What's the trick with libffi?
> The info in bug 522118 doesn't really help. (That's probably only the
> first problem I found while trying to produce FF 3.6b1.)

Rename libffi/src/x86/win32.S to sysv.S and add an underline like so,
--- win32.S     Sat Sep 26 20:07:58 2009
+++ sysv.S      Sat Oct 17 18:11:56 2009
@@ -33,7 +33,7 @@

  .text

-.globl ffi_prep_args
+.globl _ffi_prep_args

          # This assumes we are using gas.
          .balign 16

You may also need to set GREP=grep.exe and likewise for anything else
configure doesn't pick up.
Afterwards you need to run emxomf on ffi.a to create a ffi.lib.

Unluckily the libffi test suite depends on expect which depends on
tclsh. I haven't managed to build a working expect with EMX or a working
tclsh with klibc.
Dave
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Walter Meinl]

On Oct 31, 12:44 am, Dave Yeo <dave.r....@...> wrote:
> On 10/30/09 03:25 pm, Peter Weilbacher wrote:
>
> > I didn't realize we were ever unbroken. What's the trick with libffi?
> > The info in bug 522118 doesn't really help. (That's probably only the
> > first problem I found while trying to produce FF 3.6b1.)
>
I've added some workaround attachments and comments to that bug.

> You may also need to set GREP=grep.exe and likewise for anything else
> configure doesn't pick up.
I found that the problem with grep is the exe extension, though sed
awk and other exe files are found properly. I found a script in my
moztools directory for egrep and modified it and saved it as "grep"
without extension:
#!/bin/sh
exec grep ${1+"$@"}

Now libffi configure finds grep.
Walter
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Peter Weilbacher]

On 31/10/09 00:44, Dave Yeo wrote:

> Rename libffi/src/x86/win32.S to sysv.S and add an underline like so,
> --- win32.S Sat Sep 26 20:07:58 2009
> +++ sysv.S Sat Oct 17 18:11:56 2009
> @@ -33,7 +33,7 @@
>
> .text
>
> -.globl ffi_prep_args
> +.globl _ffi_prep_args
>
> # This assumes we are using gas.
> .balign 16
>
> You may also need to set GREP=grep.exe and likewise for anything else
> configure doesn't pick up.
> Afterwards you need to run emxomf on ffi.a to create a ffi.lib.

And you are doing this manually every time you build? This may be a
workaround recipe but to be able to build nightlies or even FF 3.6b1
I need a real patch that works every time (and there doesn't seem to be
a --disable-libffi).
Something like a real os2.S (copy of win32.S with this change) with
respective build patches to configure etc.

> Unluckily the libffi test suite depends on expect which depends on
> tclsh. I haven't managed to build a working expect with EMX or a working
> tclsh with klibc.

I had a working one with my old machine but on this machine it never
ran again. But Linux builds don't depend on Tcl, so this is not required
for a normal build, is it? Only when --enable-tests is on?
--
Please     | Official Warpzilla Ports: http://www.mozilla.org/ports/os2/
reply in   |
newsgroup  |          Enhanced OS/2 builds: http://pmw-warpzilla.sf.net/
     Steve's Warpzilla Tips: http://www.os2bbs.com/os2news/Warpzilla.html
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Dave Yeo-3]

On 10/31/09 01:12 am, Walter Meinl wrote:
> I found that the problem with grep is the exe extension, though sed
> awk and other exe files are found properly. I found a script in my
> moztools directory for egrep and modified it and saved it as "grep"
> without extension:
> #!/bin/sh
> exec grep ${1+"$@"}
>

set EXEEXT=.exe is set in my config.site
Dave
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Peter Weilbacher]

On 31/10/09 09:13, Peter Weilbacher wrote:

> And you are doing this manually every time you build? This may be a
> workaround recipe but to be able to build nightlies or even FF 3.6b1
> I need a real patch that works every time (and there doesn't seem to be
> a --disable-libffi).
> Something like a real os2.S (copy of win32.S with this change) with
> respective build patches to configure etc.

With Walter's latest patch (thanks!) to that bug I was finally able to
build FF 3.6b1. But all content comes up blank. I have no idea if this
is a libffi thing or caused by something else.
In the view of maybe adding back ipluginw etc. for FF 3.6 final, I built
it with GCC 3.3.5, so that could be a factor, too.
--
Please     | Official Warpzilla Ports: http://www.mozilla.org/ports/os2/
reply in   |
newsgroup  |          Enhanced OS/2 builds: http://pmw-warpzilla.sf.net/
     Steve's Warpzilla Tips: http://www.os2bbs.com/os2news/Warpzilla.html
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Dave Yeo-3]

Replying to this instead of previous message as after a crash I lost my
reply and your message no longer displays.

On 10/31/09 02:52 pm, Peter Weilbacher wrote:
> On 31/10/09 09:13, Peter Weilbacher wrote:
>
>> And you are doing this manually every time you build? This may be a
>> workaround recipe but to be able to build nightlies or even FF 3.6b1
>> I need a real patch that works every time (and there doesn't seem to be
>> a --disable-libffi).
>> Something like a real os2.S (copy of win32.S with this change) with
>> respective build patches to configure etc.

We need to run a test suite to get an idea if we're on the right path.
The libfii test suite depends on expect which I haven't got working
(only the test suite needs expect which use tcl).
Now that Dan has posted a patch to enable the ctypes unit tests we can
test with that.
This patch gets around the no rule to make fii.lib

diff --git a/js/ctypes/Makefile.in b/js/ctypes/Makefile.in
--- a/js/ctypes/Makefile.in
+++ b/js/ctypes/Makefile.in
@@ -99,6 +99,11 @@
      -Ilibffi/include \
      $(NULL)

+ifeq ($(OS_ARCH),OS2)
+libffi/.libs/ffi.lib:
+ emxomf libffi/.libs/$(LIB_PREFIX)ffi.a
+endif
+
  SHARED_LIBRARY_LIBS = \
      libffi/.libs/$(LIB_PREFIX)ffi.$(LIB_SUFFIX) \
      $(NULL)

>
> With Walter's latest patch (thanks!) to that bug I was finally able to
> build FF 3.6b1. But all content comes up blank. I have no idea if this
> is a libffi thing or caused by something else.
> In the view of maybe adding back ipluginw etc. for FF 3.6 final, I built
> it with GCC 3.3.5, so that could be a factor, too.

I doubt that it is anything to do with ctypes, seems mostly for addons.
See https://developer.mozilla.org/en/JavaScript_code_modules/ctypes.jsm.
I'm updating my tree and will test with gcc 4.4.1
Dave
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Dave Yeo-3]

On 10/31/09 02:52 pm, Peter Weilbacher wrote:
>
> With Walter's latest patch (thanks!) to that bug I was finally able to
> build FF 3.6b1. But all content comes up blank. I have no idea if this
> is a libffi thing or caused by something else.
> In the view of maybe adding back ipluginw etc. for FF 3.6 final, I built
> it with GCC 3.3.5, so that could be a factor, too.

The Firefox that I built today seems fine. Built with gcc 4.4.1 second
edition, ctypes using the win32.S patch I described and Rich's patch for
bug #525396
Dave
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Peter Weilbacher]

On 01/11/09 09:29, Dave Yeo wrote:

> On 10/31/09 02:52 pm, Peter Weilbacher wrote:
>>
>> With Walter's latest patch (thanks!) to that bug I was finally able to
>> build FF 3.6b1. But all content comes up blank. I have no idea if this
>> is a libffi thing or caused by something else.
>> In the view of maybe adding back ipluginw etc. for FF 3.6 final, I built
>> it with GCC 3.3.5, so that could be a factor, too.
>
> The Firefox that I built today seems fine. Built with gcc 4.4.1 second
> edition, ctypes using the win32.S patch I described and Rich's patch for
> bug #525396

Hmm, in the meantime I also rebuilt with 4.4.1 and about the same set of
patches you are using. Still no content displayed. What revision of
which tree did you build?
--
Please     | Official Warpzilla Ports: http://www.mozilla.org/ports/os2/
reply in   |
newsgroup  |          Enhanced OS/2 builds: http://pmw-warpzilla.sf.net/
     Steve's Warpzilla Tips: http://www.os2bbs.com/os2news/Warpzilla.html
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Walter Meinl]

On Nov 1, 12:44 pm, Peter Weilbacher <newss...@...> wrote:

> On 01/11/09 09:29, Dave Yeo wrote:
>
> > On 10/31/09 02:52 pm, Peter Weilbacher wrote:
>
> >> With Walter's latest patch (thanks!) to that bug I was finally able to
> >> build FF 3.6b1. But all content comes up blank. I have no idea if this
> >> is a libffi thing or caused by something else.
> >> In the view of maybe adding back ipluginw etc. for FF 3.6 final, I built
> >> it with GCC 3.3.5, so that could be a factor, too.
>
> > The Firefox that I built today seems fine. Built with gcc 4.4.1 second
> > edition, ctypes using the win32.S patch I described and Rich's patch for
> > bug #525396
>
> Hmm, in the meantime I also rebuilt with 4.4.1 and about the same set of
> patches you are using. Still no content displayed. What revision of
> which tree did you build?
> --
> Please     | Official Warpzilla Ports:http://www.mozilla.org/ports/os2/
> reply in   |
> newsgroup  |          Enhanced OS/2 builds:http://pmw-warpzilla.sf.net/
>      Steve's Warpzilla Tips:http://www.os2bbs.com/os2news/Warpzilla.html

I haven't build 1.9.2-branch for a longer time, but trunk does work.
Your problem sounds like
https://bugzilla.mozilla.org/show_bug.cgi?id=520894
but for that the fix was checked in about 3 weeks ago, so no clue why
it shows nothing
Walter
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Dave Yeo-3]

On 11/01/09 03:44 am, Peter Weilbacher wrote:

> On 01/11/09 09:29, Dave Yeo wrote:
>> On 10/31/09 02:52 pm, Peter Weilbacher wrote:
>>>
>>> With Walter's latest patch (thanks!) to that bug I was finally able to
>>> build FF 3.6b1. But all content comes up blank. I have no idea if this
>>> is a libffi thing or caused by something else.
>>> In the view of maybe adding back ipluginw etc. for FF 3.6 final, I built
>>> it with GCC 3.3.5, so that could be a factor, too.
>>
>> The Firefox that I built today seems fine. Built with gcc 4.4.1 second
>> edition, ctypes using the win32.S patch I described and Rich's patch for
>> bug #525396
>
> Hmm, in the meantime I also rebuilt with 4.4.1 and about the same set of
> patches you are using. Still no content displayed. What revision of
> which tree did you build?

Sorry I misread that you were having problems with trunk, which is what
I built
Dave
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Paul Smedley-5]

Hi Guys,

On Sun, 1 Nov 2009 11:44:51 UTC, Peter Weilbacher
<newsspam@...> wrote:

> On 01/11/09 09:29, Dave Yeo wrote:
> > On 10/31/09 02:52 pm, Peter Weilbacher wrote:
> >>
> >> With Walter's latest patch (thanks!) to that bug I was finally able to
> >> build FF 3.6b1. But all content comes up blank. I have no idea if this
> >> is a libffi thing or caused by something else.
> >> In the view of maybe adding back ipluginw etc. for FF 3.6 final, I built
> >> it with GCC 3.3.5, so that could be a factor, too.
> >
> > The Firefox that I built today seems fine. Built with gcc 4.4.1 second
> > edition, ctypes using the win32.S patch I described and Rich's patch for
> > bug #525396
>
> Hmm, in the meantime I also rebuilt with 4.4.1 and about the same set of
> patches you are using. Still no content displayed. What revision of
> which tree did you build?

Hmm - last I read I though most people had problems with 4.4.x and
mozilla - did I mistakenly fix those problems in 4.4.1?

--
Cheers,

Paul.
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Walter Meinl]

On Nov 1, 12:44 pm, Peter Weilbacher <newss...@...> wrote:
> On 01/11/09 09:29, Dave Yeo wrote:
>

> Hmm, in the meantime I also rebuilt with 4.4.1 and about the same set of
> patches you are using. Still no content displayed. What revision of
> which tree did you build?

Strange, I checked-out the GECKO192b1_20091014_RELBRANCH
hg identify: fff9ebf58957 (GECKO192b1_20091014_RELBRANCH)
and can still apply the attachment of
https://bugzilla.mozilla.org/show_bug.cgi?id=520894 (checkin
b90714b554ed)
also your checkins f300d0e3b35b (print preview), 3ac6c2063f63
(nsSound), c16794559aed (plugin focus)
seem not to be pushed to the release branch.
Walter
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2

[Peter Weilbacher]

On 01/11/09 21:58, Paul Smedley wrote:
> On Sun, 1 Nov 2009 11:44:51 UTC, Peter Weilbacher
> <newsspam@...>  wrote:
>> Hmm, in the meantime I also rebuilt with 4.4.1 and about the same set of
>> patches you are using. Still no content displayed. What revision of
>> which tree did you build?
>
> Hmm - last I read I though most people had problems with 4.4.x and
> mozilla - did I mistakenly fix those problems in 4.4.1?

IIRC it was only the second release of 4.4.0 (dated 2009-05-03?) that
was problematic. I think 4.4.1 always worked here.

(What I still would like to see if you have an hour or so, is a build of
GCC 3.4.6 with the change that made "-static-libgcc" work. Then we could
get rid of most ugly patches for 3.3.5 _and_ get working Java plugin
support. I wanted to look at that, too, but it seems I wasn't desperate
enough yet.)
--
Please     | Official Warpzilla Ports: http://www.mozilla.org/ports/os2/
reply in   |
newsgroup  |          Enhanced OS/2 builds: http://pmw-warpzilla.sf.net/
     Steve's Warpzilla Tips: http://www.os2bbs.com/os2news/Warpzilla.html
_______________________________________________
dev-ports-os2 mailing list
dev-ports-os2@...
https://lists.mozilla.org/listinfo/dev-ports-os2