Bug 199539
| Summary: | math/plplot: upgrading to 5.11.0 | ||
|---|---|---|---|
| Product: | Ports & Packages | Reporter: | Thierry Thomas <thierry> |
| Component: | Individual Port(s) | Assignee: | Thierry Thomas <thierry> |
| Status: | Closed FIXED | ||
| Severity: | Affects Some People | CC: | adamw, marino, thierry, woodsb02 |
| Priority: | --- | Keywords: | patch-ready |
| Version: | Latest | ||
| Hardware: | Any | ||
| OS: | Any | ||
| Attachments: | |||
|
Description
Thierry Thomas
2015-04-19 11:15:28 UTC
Created attachment 155771 [details] Patch to update math/plplot to 5.11.0 Thierry, you and I must be on the same wavelength... I have been working on this patch for a lot of the weekend, independent of you. Talk about unnecessary overlap! However, our patches have a LOT of similarities! I have attached my patch, which builds successfully with poudriere testport with all options disabled and all options enabled. Differences with my patch: - Convert to new options helpers - iwidgets is only BUILD_DEPENDS - No issues with tcl/tk/itcl/itk when building as you reported - wxgtk 2.6 or 2.8 can be used (http://sourceforge.net/p/plplot/wiki/WxWidgets/) - Removed PLD_wxpng build flag (http://sourceforge.net/p/plplot/mailman/message/33915239/) - Remove LUA_MODLIBDIR as this isn't required and breaks plist checks - Remove patches which now only modify source code comments (other diffs upstreamed) - take maintainership Thoughts? Created attachment 155772 [details]
Successful poudriere testport on FreeBSD 10.1 amd64 with all options unset
Created attachment 155773 [details]
Successful poudriere testport on FreeBSD 10.1 amd64 with all options set
Comment on attachment 155725 [details]
Preliminary version of patch from plplot-5.10.0 to plplot-5.11.0
The patch submitted by Ben Woods is better (it solves the itk / itcl problem).
Thanks Ben, your patch is great! But we cannot commit it yet, because science/gnudatalanguage and math/plplot-ada must support this new version (marino added to Cc: list for plplot-ada). Ben, maybe you can supply a poudriere log for plplot-ada using the new plplot? If we get luckily it will build out of the box. poudriere testport of math/plplot-ada fails with the following errors: =======================<phase: configure >============================ ===> plplot-ada-5.11.0 depends on file: /usr/local/gcc-aux/bin/ada - found ===> plplot-ada-5.11.0 depends on shared library: libplplotd.so - not found ===> Verifying for libplplotd.so in /usr/ports/math/plplot ===> Installing existing package /packages/All/plplot-5.11.0.txz pkg-static: Insufficient privileges to add packages *** Error code 77 I am currently trying to understand what is causing this error - any help would be appreciated. I do not see this error when performing poudriere testport of math/plplot-ada without having made my changes to math/plplot... so this seems to be influenced by this patch. I am running an older poudriere on freebsd 10, I suppose I could try it and see if I see the same problem. I would think this is a pkg or poudriere bug. I guess that the problem is libplplotd.so being renamed to libplplot.so (without 'D'). (In reply to Thierry Thomas from comment #9) ah, right. it's breaking the dependency specification. Ben, given that, come up with a patch that: 1) fixes the dependency 2) bumps PORTREVISION If that allows adaplot to build, then attach the patch the PR. both will need to be committed simultaneously. (In reply to Thierry Thomas from comment #9) > I guess that the problem is libplplotd.so being renamed to libplplot.so (without 'D'). You are spot on Thierry - thanks for the heads up... I had spent too long looking through poudriere and pkg code to try and find a bug there! I'll submit patches for math/plplot-ada and science/gnudatalanguage Created attachment 155822 [details]
Patch math/plplot-ada to fix LIB_DEPENDS on plplot.so and bump PORTREVISION
Created attachment 155823 [details]
Successful poudriere testport on FreeBSD10.1 amd64 for math/plplot-ada
Comment on attachment 155822 [details]
Patch math/plplot-ada to fix LIB_DEPENDS on plplot.so and bump PORTREVISION
approved
Created attachment 155842 [details] Patch science/gnudatalanguage to fix build with plplot 5.11.0 - Change LIB_DEPENDS from libplplotd.so to libplplot.so - Add patch-CMakeModules_FindPlplot.cmake to remove 'd' from library names in search - Add patch-src_plotting__contour.cpp to change plstream::fill/tr to plcallback::fill/tr as per note 1.3 in plplot-5.11.0 release notes: http://sourceforge.net/p/plplot/plplot/ci/plplot-5.11.0/tree/README.release 1.3 Backwards-incompatible change to our C++ API The plstream methods fill, tr0, tr1, tr2, and tr2p have been changed, fixing a bug where the correct stream was not selected. This means these functions are no longer static and can no longer be used as callbacks in the shades method and its variants, the cont method, the imagefr method, and the vect method. Instead a plcallback namespace has been created which includes these methods as C-style functions. Code which previously used the plstream methods above will need to be recompiled and rebuilt using the plcallback versions (i.e., replace all plstream::fill callback use by plcallback::fill and similarly for tr0, tr1, tr2, and tr2p). Please see the documentation for a full description. Created attachment 155843 [details]
Successful poudriere testport on FreeBSD10.1 amd64 for science/gnudatalanguage
Patch for science/gnudatalanguage approved! Committed, thanks! A commit references this bug: Author: thierry Date: Wed Apr 22 20:16:11 UTC 2015 New revision: 384520 URL: https://svnweb.freebsd.org/changeset/ports/384520 Log: - Upgrade math/plplot to 5.11.0 Announcement at <https://sourceforge.net/p/plplot/plplot/ci/plplot-5.11.0/tree/README.release> - Take maintainership - Chase dependencies. PR: ports/199539 Submitted by: Ben Woods <woodsb02 (at) gmail.com> Approved by: marino (for plplot-ada), /me (for gnudatalanguage) Changes: head/math/plplot/Makefile head/math/plplot/distinfo head/math/plplot/files/patch-bindings__tk-x-plat__plolotter.c head/math/plplot/files/patch-bindings__tk__plframe.c head/math/plplot/files/patch-bindings__tk__plserver.c head/math/plplot/files/patch-bindings__tk__tkMain.c head/math/plplot/files/patch-cmake_modules_FindVGA.cmake head/math/plplot/files/patch-cmake_modules_csiro.cmake head/math/plplot/files/patch-cmake_modules_instdirs.cmake head/math/plplot/files/patch-cmake_modules_linuxvga.cmake head/math/plplot/files/patch-cmake_modules_pkg-config.cmake head/math/plplot/files/patch-doc__docbook__src__tcl.xml head/math/plplot/files/patch-examples__python__CMakeLists.txt head/math/plplot/files/patch-examples__tk__xtk01.c head/math/plplot/files/patch-examples__tk__xtk02.c head/math/plplot/files/patch-examples__tk__xtk04.c head/math/plplot/files/patch-include_plConfig.h head/math/plplot/files/patch-include_plConfig.h.in head/math/plplot/files/patch-plplot_test__CMakeLists.txt head/math/plplot/files/patch-scripts__CMakeLists.txt head/math/plplot/files/patch-util__plrender.c head/math/plplot/pkg-plist head/math/plplot-ada/Makefile head/science/gnudatalanguage/Makefile head/science/gnudatalanguage/files/patch-CMakeLists.txt head/science/gnudatalanguage/files/patch-CMakeModules_FindPlplot.cmake head/science/gnudatalanguage/files/patch-CMakeModules_Findlibps.cmake head/science/gnudatalanguage/files/patch-src__base64.hpp head/science/gnudatalanguage/files/patch-src__basic_fun.cpp head/science/gnudatalanguage/files/patch-src__gshhs.cpp head/science/gnudatalanguage/files/patch-src__gsl_fun.cpp head/science/gnudatalanguage/files/patch-src__str.cpp head/science/gnudatalanguage/files/patch-src_plotting__contour.cpp Heh, I forgot that plplot-ada PORTVERSION is always the same as the main plplot version, so the PORTREVISION bump was totally unnecessary. I was looking at it because plplot-ada has been updated upstream (so I would need to bump portrevision anyway). It just means that it comes in at "2" rather than the normal "0" ah, I think portscout confused me. The bindings must be within plplot itself, so there is nothing more to update. When portscout saw the plplot new version available, then it alerted me that plplot-ada also had a new version available. So really the update is done, just with an extra PORTREVISION bump. sorry, I forgot how this port worked. All good, guess we can take out the bonus portrevision when we next update plplot. |