The short story is I have decided to put off the release to early next week (if a lot of progress is made over the weekend) or even later in the week.
Now here is the longer story about where we stand with this release. @Arjen: I very much appreciate the guidance you have given about the problems Phil has been finding. Here are the Windows issues as I understand them now from what Arjen has said, but please correct this summary if I get something wrong. (1) The "Visual Studio 11" generator (for the VS2012 IDE) is not working correctly. This is likely (see below for my own conclusions confirming that) an inherent CMake bug and nothing to do with PLplot. So according to Arjen's advice we note the issue but do not delay the release because of it. After all, it is not a showstopper for individuals with VS2012 since nmake should always be available to them for building PLplot. (2) Subsequently, Arjen found similar but less severe issues with the "Visual Studio 10" generator (for the VS2010 IDE), but I assume we should treat this the same as (1). (3) Phil has found the following issue which I spotted in the CMake output that he sent for the "Visual Studio 11" generator: CMake Error at cmake/modules/plplot_functions.cmake:210 (list): list sub-command REMOVE_DUPLICATES requires list to be present. Call Stack (most recent call first): cmake/modules/shapelib.cmake:34 (filter_rpath) cmake/modules/plplot.cmake:513 (include) CMakeLists.txt:111 (include) That error message implies none of CMAKE_C_IMPLICIT_LINK_DIRECTORIES, CMAKE_CXX_IMPLICIT_LINK_DIRECTORIES, or CMAKE_FORTRAN_IMPLICIT_LINK_DIRECTORIES are defined which is impossible (!) according to the CMake documentation. My conclusion from this is the "Visual Studio 11" generator is severely broken. Neverthless, I did make the filter_rpath command robust against this unexpected bad result for those variables in revision 12864. (4) Phil has found a really strange -L+ issue where the workaround is to set -DENABLE_d=OFF. I assume that issue is also for the "Visual Studio 11" generator. @Phil: I strongly second Arjen's suggestion that you move to nmake for now. Once we have that working perfectly for you, then we can go back (post-release) to try and figure out what is wrong with VS2012 (which will likely involve CMake bug reports rather than anything to do with PLplot). If you agree this is the right approach, would you please try the "NMake Makefiles" generator as soon as possible, and let us know if (a) there are any cmake issues left for that case, and (b) assuming all is well with the cmake results, does the PLplot build with nmake work and pass all your usual tests? Of course, there are still some uncertainties about the Windows platform until Phil responds to the above request, but my guess from Arjen's good experiences with nmake is that it will work well for Phil as well. IMPORTANT. But if Phil does find nmake issues for his platform, I would view those as release critical so I would be willing to delay the release until you guys figure out whatever that (hypothetical) problem is. As far as I know there are no release-critical Mac OS X issues. Jerry did find a general issue with the test logic that I recently fixed, but my understanding from what he has said is everything else has been working well for him on that platform. Here is the summary of the remaining release-critical issues for the Linux platform (including the reason I have decided to delay the release for at least a few days). I discovered these issues as a result of the option for comprehensive testing of PLplot I enabled last night for the epa_build project. (1) A potential showstopper was segfaults for -dev tk for the combination of Tcl/Tk8.6 and using the stubs version of the Tcl/Tk libraries. I have worked around this issue for this release by switching back to using the ordinary Tcl/Tk libraries by default (the approach we had been using until last month). But using the stubs versions of the Tcl/Tk libraries is highly recommended by the Tcl/Tk developers. They also strongly suggest (as do I) moving from the deprecated Tcl/Tk API we currently use to the recommended Tcl/Tk API. That may solve the segfault issue, but there are obviously a lot of high-priority changes we need to make in the Tcl/Tk bindings and examples post-release. (2) There is a build-system inconsistency about the pkg-config version that is being used. I hope to address that today. (3) The real issue with a release this weekend (assuming nmake is fine for Phil) is I am suddenly at least a couple of days behind the schedule I worked out for this release. For example, I still have plans to do one more complete round of testing via epa_build on Linux and also on MinGW/MSYS/Wine, and there are bound to be some issues to fix especially for the latter platform. So the start of the release process which I had planned for Thursday morning is likely only going to happen Monday morning at the earliest. In sum, there is still some open-ended stuff to do for this release on Linux, MinGW/MSYS/Wine and also possibly MSVC/nmake platforms so it is hard to predict exact timing, but my guess is it going to be at least Monday morning before I start the actual release process and at least Tuesday before I finish that release process. But I will keep you guys informed as the currently large uncertainties gradually disappear about when the release will actually happen. I also am taking this opportunity to publicly thank everybody who has been so heavily testing PLplot these last few days. The fixes that have resulted from that testing have already added greatly to the quality of this forthcoming release! Alan __________________________ Alan W. Irwin Astronomical research affiliation with Department of Physics and Astronomy, University of Victoria (astrowww.phys.uvic.ca). Programming affiliations with the FreeEOS equation-of-state implementation for stellar interiors (freeeos.sf.net); the Time Ephemerides project (timeephem.sf.net); PLplot scientific plotting software package (plplot.sf.net); the libLASi project (unifont.org/lasi); the Loads of Linux Links project (loll.sf.net); and the Linux Brochure Project (lbproject.sf.net). __________________________ Linux-powered Science __________________________ ------------------------------------------------------------------------------ Rapidly troubleshoot problems before they affect your business. Most IT organizations don't have a clear picture of how application performance affects their revenue. With AppDynamics, you get 100% visibility into your Java,.NET, & PHP application. Start your 15-day FREE TRIAL of AppDynamics Pro! http://pubads.g.doubleclick.net/gampad/clk?id=84349831&iu=/4140/ostg.clktrk _______________________________________________ Plplot-devel mailing list [email protected] https://lists.sourceforge.net/lists/listinfo/plplot-devel
