Re: [Aqsis-commits] [SCM] Aqsis Renderer branch, master, updated. CVS_Tip-2538-g5ba288b
|
View:
New views
10 Messages
—
Rating Filter:
Alert me
|
 |
|
|
|
Re: [Aqsis-commits] [SCM] Aqsis Renderer branch, master, updated. CVS_Tip-2538-g5ba288b
by Chris Foster-5
::
Rate this Message:
On Tue, Sep 8, 2009 at 7:37 AM, Chris Foster<chris42f@...> wrote:
> This patch converts the entire file ri.cpp to dos newline characters, > which makes for a massively pointless diff and completely confuses "git > blame". I'm not sure how best to avoid this at your end, but it's worth > thinking about because it's quite disruptive to the history :-( Hmm, there is a git config option: git config core.autocrlf true Can you use that? ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Aqsis-development mailing list Aqsis-development@... https://lists.sourceforge.net/lists/listinfo/aqsis-development |
|
|
Re: [Aqsis-commits] [SCM] Aqsis Renderer branch, master, updated. CVS_Tip-2538-g5ba288b
by Paul Gregory-4
::
Rate this Message:
I've found if I enable that I end up with tonnes of false modifications on my end :(
Paul 2009/9/7 Chris Foster <chris42f@...>
------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Aqsis-development mailing list Aqsis-development@... https://lists.sourceforge.net/lists/listinfo/aqsis-development |
|
|
Re: [Aqsis-commits] [SCM] Aqsis Renderer branch, master, updated. CVS_Tip-2538-g5ba288b
by Chris Foster-5
::
Rate this Message:
On Tue, Sep 8, 2009 at 7:46 AM, Paul Gregory<pgregory@...> wrote:
> I've found if I enable that I end up with tonnes of false modifications on > my end :( Hmm, odd. The following seems to suggest that it can do some weird things, but works ok when setup: http://github.com/guides/dealing-with-newlines-in-git Or is it that we have some files with mixed or pure CRLF line endings in the repo? We shouldn't really. Luckily it seems to be reasonably easy to fix the problem with the ri.cpp history using a branch and merge, so I've done that now and committed. ~Chris ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Aqsis-development mailing list Aqsis-development@... https://lists.sourceforge.net/lists/listinfo/aqsis-development |
|
|
Re: [Aqsis-commits] [SCM] Aqsis Renderer branch, master, updated. CVS_Tip-2538-g5ba288b
by Matthias Baas-3
::
Rate this Message:
Hi,
Paul Gregory wrote: > I've found if I enable that I end up with tonnes of false modifications > on my end :( I have just moved the cgkit sources over to git as well and was facing the same problem. On Windows, I actually did set the core.autocrlf and core.safecrlf variables which seems to work fine. I also noticed that on some files, git reported changes even though I didn't modify anything. This happens when the files are stored in the repository using CRLF line endings (because the above options convert this back to LF when committing, so the files in your local checkout do differ from what will be in the repository). I just committed these changes back so that the repository is in a consistent LF-only state. From then on, things went fine so far. (in my case, it was probably a minor mistake to set those variables on the local repository after I cloned the code (instead of setting them globally). The thing is, git only reports these line ending changes when you touch the offending files in some way (either modifying them or checking them out explicitly again) which means every now and then I discover a new file that still has CRLF line endings and needs conversion) - Matthias - ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Aqsis-development mailing list Aqsis-development@... https://lists.sourceforge.net/lists/listinfo/aqsis-development |
|
|
Re: [Aqsis-commits] [SCM] Aqsis Renderer branch, master, updated. CVS_Tip-2538-g5ba288b
by Chris Foster-5
::
Rate this Message:
On Tue, Sep 8, 2009 at 9:09 AM, Matthias Baas<matthias.baas@...> wrote:
> (in my case, it was probably a minor mistake to set those variables on > the local repository after I cloned the code (instead of setting them > globally). The thing is, git only reports these line ending changes when > you touch the offending files in some way (either modifying them or > checking them out explicitly again) which means every now and then I > discover a new file that still has CRLF line endings and needs conversion) The page I just linked seems to suggest that all the files can easily be converted at once by deleting the working tree entirely (all except the .git directory), making sure the autocrlf is set and then doing a git reset --hard HEAD and committing any changes that result... ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Aqsis-development mailing list Aqsis-development@... https://lists.sourceforge.net/lists/listinfo/aqsis-development |
|
|
Re: [Aqsis-commits] [SCM] Aqsis Renderer branch, master, updated. CVS_Tip-2538-g5ba288b
by Paul Gregory-4
::
Rate this Message:
Ok, I tried that...
On Windows with Cygwin git, I get nothing reported from git status after doing that with core.autocrlf set to true. On Linux, I get... # modified: release-notes/1.4/summary-1.4.txt # modified: tools/displays/sdcBMP/d_sdcBMP.dsp # modified: tools/eqsl/eqsl.fl # modified: tools/integration/BtoR/BtoR.py # modified: tools/integration/BtoR/BtoRAdapterClasses.py # modified: tools/integration/BtoR/BtoRGUIClasses.py # modified: tools/integration/BtoR/BtoRMain.py # modified: tools/integration/BtoR/BtoRSpaceHandler.py # modified: tools/integration/BtoR/BtoRTypes.py # modified: tools/integration/BtoR/README.GUI # modified: tools/integration/BtoR/UITest.py # modified: tools/piqsl/piqsl_ui.fl all of which are line ending differences. So which is it? Paul 2009/9/8 Chris Foster <chris42f@...>
------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Aqsis-development mailing list Aqsis-development@... https://lists.sourceforge.net/lists/listinfo/aqsis-development |
|
|
Re: [Aqsis-commits] [SCM] Aqsis Renderer branch, master, updated. CVS_Tip-2538-g5ba288b
by Paul Gregory-4
::
Rate this Message:
What's more, I can't actually do anything on Linux with that, as I can't rebase to get the latest changes because I have local changes, I can't stash them because doing so will checkout the latest state, which of course applies the lineending filter, resulting in the same state, continue ad infinitum.
I swear, whoever it was that made the decision to not standardise line-endings needs a good hard slap. Paul 2009/9/8 Paul Gregory <pgregory@...> Ok, I tried that... ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Aqsis-development mailing list Aqsis-development@... https://lists.sourceforge.net/lists/listinfo/aqsis-development |
|
|
Re: [Aqsis-commits] [SCM] Aqsis Renderer branch, master, updated. CVS_Tip-2538-g5ba288b
by Matthias Baas-3
::
Rate this Message:
Paul Gregory wrote:
> Ok, I tried that... > > On Windows with Cygwin git, I get nothing reported from git status after > doing that with core.autocrlf set to true. Where did you try it out? Could it be that it was within another directory tracked by git? That's what I did first (I was inside a tmp directory within my cgkit sources) and then I also got a clean status. When I tried again outside a git directory, it looked like this: C:\tmp>git clone git://aqsis.git.sourceforge.net/gitroot/aqsis/aqsis Initialized empty Git repository in /cygdrive/c/tmp/aqsis/.git/ remote: Counting objects: 48183, done. remote: Compressing objects: 100% (15640/15640), done. remote: Total 48183 (delta 33943), reused 45553 (delta 31978) Receiving objects: 100% (48183/48183), 46.54 MiB | 1156 KiB/s, done. Resolving deltas: 100% (33943/33943), done. Checking out files: 100% (827/827), done. C:\tmp>cd aqsis C:\tmp\aqsis>git status # On branch master nothing to commit (working directory clean) >>>>>>>> now I delete everything except the .git directory C:\home\Sources\tmp\aqsis>git config core.autocrlf true C:\home\Sources\tmp\aqsis>git config core.safecrlf true C:\home\Sources\tmp\aqsis>git reset --hard HEAD Checking out files: 100% (827/827), done. HEAD is now at c0e54c4 Bugfix 2853236: No AOV for volume shaders C:\tmp\aqsis>git status # On branch master # Changed but not updated: # (use "git add <file>..." to update what will be committed) # (use "git checkout -- <file>..." to discard changes in working directory) # # modified: tools/displays/sdcBMP/d_sdcBMP.dsp # modified: tools/eqsl/eqsl.fl # modified: tools/integration/BtoR/BtoR.py # modified: tools/integration/BtoR/BtoRAdapterClasses.py # modified: tools/integration/BtoR/BtoRGUIClasses.py # modified: tools/integration/BtoR/BtoRMain.py # modified: tools/integration/BtoR/BtoRSpaceHandler.py # modified: tools/integration/BtoR/BtoRTypes.py # modified: tools/integration/BtoR/README.GUI # modified: tools/integration/BtoR/UITest.py # modified: tools/piqsl/piqsl_ui.fl # no changes added to commit (use "git add" and/or "git commit -a") As I said, when I do the exact same thing in that tmp directory within my cgkit sources, the second git status is clean (?!?). The other funny thing is that git seems to handle *.txt files differently as the summary-1.4.txt file is not reported. By the way, I can confirm that these files do have DOS line endings in the repository (so Linux people should see the DOS line endings as well). > On Linux, I get... Are you using autocrlf on Linux as well? (I'm not using it on OSX as I don't expect I will add files with DOS line endings when I'm not under Windows. And apart from that, the git documentation sounds as if it would always convert from LF to CRLF when checking files out (instead of converting to native line endings) which doesn't make sense on non-Windows systems anyway, does it? (but I didn't test this and I'm currently under Windows and won't reboot just because of that... :) )) > # modified: release-notes/1.4/summary-1.4.txt > # modified: tools/displays/sdcBMP/d_sdcBMP.dsp > # modified: tools/eqsl/eqsl.fl > # modified: tools/integration/BtoR/BtoR.py > # modified: tools/integration/BtoR/BtoRAdapterClasses.py > # modified: tools/integration/BtoR/BtoRGUIClasses.py > # modified: tools/integration/BtoR/BtoRMain.py > # modified: tools/integration/BtoR/BtoRSpaceHandler.py > # modified: tools/integration/BtoR/BtoRTypes.py > # modified: tools/integration/BtoR/README.GUI > # modified: tools/integration/BtoR/UITest.py > # modified: tools/piqsl/piqsl_ui.fl - Matthias - ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Aqsis-development mailing list Aqsis-development@... https://lists.sourceforge.net/lists/listinfo/aqsis-development |
|
|
Re: [Aqsis-commits] [SCM] Aqsis Renderer branch, master, updated. CVS_Tip-2538-g5ba288b
by Chris Foster-5
::
Rate this Message:
Hey guys,
For those not on IRC, we chatted about the git issues this morning... http://chat.aqsis.org/index.n?year=2009&month=8&day=8 and I'm starting to conclude that the confusion could be partly due to a git bug, or at the least, git is behaving in a very odd manner. Here's what I'm seeing: First clone the git repo: git clone git://aqsis.git.sourceforge.net/gitroot/aqsis/aqsis nltest_src cd nltest_src set the autocrlf option to 'input' (I gather this is the appropriate setting for linux, where we want LF for end of line rather than CRLF in the working tree): git config core.autocrlf input add Now remove the working tree as stated on the github page, and do a git reset followed by git status to see what the state of the tree is: rm -Rf * && git reset --hard HEAD && git status Now the odd thing happens. Sometimes I see a big list of modified files like the following which contains pretty much all the files with CRLF line endings. This is what I would expect, based on some messing around with the dos2unix utility: # modified: cmake/modules/FindNSIS.cmake # modified: cmake/modules/FindOpenEXR.cmake # modified: cmake/platform/windows.cmake # modified: distribution/win/info.rc.in.cmake # modified: distribution/win/nsis/setup.nsi.in.cmake # modified: examples/features/archives/bike.rib # modified: examples/features/archives/render.bat # modified: examples/features/bake/render.bat # modified: examples/features/blobby/air/blobplane.rib # modified: examples/features/curves/bezier.rib # modified: examples/features/curves/render.bat # modified: examples/features/levelofdetail/detail.rib # modified: examples/features/levelofdetail/render.bat # modified: examples/features/motionblur/camera.rib # modified: examples/features/motionblur/deformation.rib # modified: examples/features/motionblur/render_camera.bat # modified: examples/features/motionblur/render_deformation.bat # modified: examples/features/multipass/aov.rib # modified: examples/features/multipass/myval.sl # modified: examples/features/multipass/render.bat # modified: examples/features/objectinstance/render.bat # modified: examples/features/objectinstance/singlepolygon.rib # modified: examples/features/occlusion/envlight.sl # modified: examples/features/occlusion/occlmap.py # modified: examples/features/occlusion/occlmap.rib # modified: examples/features/occlusion/render.bat # modified: examples/features/occlusion/simple.rib # modified: examples/features/occlusion/world.rib # modified: examples/features/pointcloud/render.bat # modified: examples/features/pointcloud/simple.rib # modified: examples/features/pointcloud/simple_texture3d.rib # modified: examples/features/shadows/autoshadow.rib # modified: examples/features/shadows/render_autoshadow.bat # modified: examples/features/shadows/render_softshadow.bat # modified: examples/features/shadows/softshadow.rib # modified: examples/features/solidmodeling/csg.rib # modified: examples/features/solidmodeling/render.bat # modified: examples/features/subdivision/render.bat # modified: examples/features/textures/render.bat # modified: examples/features/textures/sticky.rib # modified: examples/procedurals/menger/render.bat # modified: examples/scenes/microbe/render.bat # modified: include/aqsis/ri/pointcloud.h # modified: libs/build_tools/boostautotestmain.cpp # modified: libs/shadervm/shaderexecenv/shadeops_bake3d.cpp # modified: release-notes/1.4/summary-1.4.txt # modified: tools/displays/sdcBMP/d_sdcBMP.dsp # modified: tools/eqsl/eqsl.fl # modified: tools/integration/BtoR/BtoR.py # modified: tools/integration/BtoR/BtoRAdapterClasses.py # modified: tools/integration/BtoR/BtoRGUIClasses.py # modified: tools/integration/BtoR/BtoRMain.py # modified: tools/integration/BtoR/BtoRSpaceHandler.py # modified: tools/integration/BtoR/BtoRTypes.py # modified: tools/integration/BtoR/README.GUI # modified: tools/integration/BtoR/UITest.py # modified: tools/piqsl/piqsl_ui.fl However, sometimes git status reports *much* smaller lists: # modified: release-notes/1.4/summary-1.4.txt # modified: tools/displays/sdcBMP/d_sdcBMP.dsp # modified: tools/eqsl/eqsl.fl # modified: tools/integration/BtoR/BtoR.py # modified: tools/integration/BtoR/BtoRAdapterClasses.py # modified: tools/integration/BtoR/BtoRGUIClasses.py # modified: tools/integration/BtoR/BtoRMain.py # modified: tools/integration/BtoR/BtoRSpaceHandler.py # modified: tools/integration/BtoR/BtoRTypes.py # modified: tools/integration/BtoR/README.GUI # modified: tools/integration/BtoR/UITest.py # modified: tools/piqsl/piqsl_ui.fl after some googling, this appears to be related to (possibly due to?) the "racy git" problem, as mentioned here: http://osdir.com/ml/git/2009-05/msg01147.html The only way I've found to make things work reliably is to force git to regenerate the index file by making it think that the index is out of date compared to the working tree files: rm -Rf * && git reset --hard HEAD && touch -t 01010000 .git/index && git status (the timestamp 01010000 corresponds to 1st Jan 00:00 this year) So the oddness we were observing this morning is probably nothing to do with git's heuristic for determining whether a file is text or binary (phew). Rather, it's probably related to an obscure race condition between the modification times of the files in the tree and the index file. ~Chris. ------------------------------------------------------------------------------ Let Crystal Reports handle the reporting - Free Crystal Reports 2008 30-Day trial. Simplify your report design, integration and deployment - and focus on what you do best, core application coding. Discover what's new with Crystal Reports now. http://p.sf.net/sfu/bobj-july _______________________________________________ Aqsis-development mailing list Aqsis-development@... https://lists.sourceforge.net/lists/listinfo/aqsis-development |
| Free embeddable forum powered by Nabble | Forum Help |
