From: J. L. M. <jl...@im...> - 2015-04-17 15:50:43
|
Hello! When configuring PLplot 5.11.0 from pkgsrc using my own pkgsrc plplot package definition on OS X Yosemite using CMake 3.0.2, I get the following error: === CMake Error at cmake/modules/pkg-config.cmake:202 (string): string sub-command REGEX, mode REPLACE needs at least 6 arguments total to command. Call Stack (most recent call first): cmake/modules/pkg-config.cmake:355 (pkg_config_link_flags) bindings/c++/CMakeLists.txt:83 (pkg_config_file) === I think the problem is that the ${link_flags} argument is not quoted, and if it's empty, then Cmake doesn't see it as an argument. If I quote it, then the configure succeeds. Below is a patch against the 5.11.0 release. Thank you! Lewis --- cmake/modules/pkg-config.cmake.orig 2015-04-12 10:08:04.000000000 +0000 +++ cmake/modules/pkg-config.cmake 2015-04-17 14:47:38.000000000 +0000 @@ -204,7 +204,7 @@ "/System/Library/Frameworks/([^ ]*)\\.framework" "-framework \\1" link_flags - ${link_flags} + "${link_flags}" ) #message("(frameworks) link_flags = ${link_flags}") endif(CMAKE_SYSTEM_NAME STREQUAL "Darwin") |
From: Alan W. I. <ir...@be...> - 2015-05-18 22:01:29
|
On 2015-04-17 10:31-0500 J. Lewis Muir wrote: > Hello! > > When configuring PLplot 5.11.0 from pkgsrc using my own pkgsrc plplot > package definition on OS X Yosemite using CMake 3.0.2, I get the > following error: > > === > CMake Error at cmake/modules/pkg-config.cmake:202 (string): > string sub-command REGEX, mode REPLACE needs at least 6 arguments total to > command. > Call Stack (most recent call first): > cmake/modules/pkg-config.cmake:355 (pkg_config_link_flags) > bindings/c++/CMakeLists.txt:83 (pkg_config_file) > === Hi Lewis: We prefer issues first to be discussed either here or on plplot-general (because the issues often turn out not to be bugs or can be fixed immediately with no bug triage or have already been fixed in our git version). Then as a last resort post the issue to the bug tracker. Thanks very much for following that preferred reporting model. But somehow that model broke down and the report you made above just slipped through here with no response from anybody so you did have to resort to using the bug tracker. Sorry about that! I am pretty sure I have already fixed the above problem as part of my large recent effort to make the pkg-config linking model analogous to the much better debugged CMake linking model for PLplot. So could you give the latest version of our git master branch (which will soon form the basis for the forthcoming 5.11.1 release) a try to see if that solves the above issue? If so I will close the corresponding bug report that you just made at <http://sourceforge.net/p/plplot/bugs/163/>. 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 __________________________ |
From: Alan W. I. <ir...@be...> - 2015-05-22 03:13:41
|
Hi Lewis: On 2015-05-21 17:51-0500 J. Lewis Muir wrote: > On 5/18/15 5:01 PM, Alan W. Irwin wrote: >> So could you give the latest version of our git master branch >> (which will soon form the basis for the forthcoming 5.11.1 >> release) a try to see if that solves the above issue? If so I >> will close the corresponding bug report that you just made at >> <http://sourceforge.net/p/plplot/bugs/163/>. > > Hi, Alan. > > I just tried, and indeed the problem I reported no longer occurs in the > git master branch. Good. I have therefore changed the status of <http://sourceforge.net/p/plplot/bugs/163/> to "closed-fixed". 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 __________________________ |
From: Alan W. I. <ir...@be...> - 2015-05-22 03:21:37
|
Hi Lewis: I am going to use a new subject line for this question for obvious reasons. On 2015-05-21 17:51-0500 J. Lewis Muir wrote: > P.S. Even though the problem I reported appears to have been fixed, > my build did not succeed, so I'm guessing I hit a new issue. The > build failed as indicated below. Unfortunately, I don't have time to > investigate, so this is really just an FYI. If you have people testing > on OS X and things work for them, then it may be something related to my > setup or building in pkgsrc. Or maybe you don't have people testing on > OS X, and this is something that you'll eventually run into on OS X at > some point before you make an official 5.11.1 release. This is on OS X > Yosemite 10.10.3. The fact is we have suffered for a very long time from a lack of testing on the OS X platform so some build bugs may be currently present on that platform. However, if those exist it should be absolutely straightforward to fix them once we know exactly what they are. I do know some of our developers here have access to that platform so I hope your report inspires them to do some testing on that platform starting with using the VERBOSE=1 option for the make command to find out exactly how our CMake-based build system instructions translate to actual build commands on OS X. If some developer here with OS X build expertise is interested in testing on that platform, my additional advice is to look at our current override file which says (based on Mac OS X knowledge from nearly a decade ago!) software@raven> cat cmake/UserOverride.cmake # User overrides of CMake default compiler and linking rules. if(APPLE) set(CMAKE_SHARED_LINKER_FLAGS "-single_module" CACHE STRING "Flags used to link a shared library.") endif(APPLE) It is possible, for example, that that -single_module override may be messing up modern Mac OS X builds with CMake and we should just completely get rid of cmake/UserOverride.cmake. But it takes someone with Mac OS X build expertise to advise us on that question. Alan > > ~~ > Linking C shared module svg.so > [ 88%] Building C object drivers/CMakeFiles/xwin.dir/xwin.c.o > [ 88%] Built target xfig > Scanning dependencies of target pltek > [ 88%] Built target test-drv-info > [ 88%] Built target svg > [ 90%] Building C object utils/CMakeFiles/pltek.dir/pltek.c.o > Scanning dependencies of target test_null_dyndriver > Scanning dependencies of target test_mem_dyndriver > [ 93%] Generating test_dyndrivers_dir/null.driver_info > [ 93%] Generating test_dyndrivers_dir/mem.driver_info > dyld: Library not loaded: @rpath/libplplot.13.dylib > Referenced from: /home/pbulk/build/local/plplot/work/plplot-5.11.1/drivers/./test-drv-info > Reason: image not found > dyld: Library not loaded: @rpath/libplplot.13.dylib > Referenced from: /home/pbulk/build/local/plplot/work/plplot-5.11.1/drivers/./test-drv-info > Reason: image not found > sh: line 6: 95551 Trace/BPT trap: 5 ./test-drv-info null > /home/pbulk/build/local/plplot/work/plplot-5.11.1/drivers/test_dyndrivers_dir/null.driver_info > --- drivers/test_dyndrivers_dir/null.driver_info --- > *** [drivers/test_dyndrivers_dir/null.driver_info] Error code 133 > > bmake: stopped in /home/pbulk/build/local/plplot/work/plplot-5.11.1 > 1 error > > bmake: stopped in /home/pbulk/build/local/plplot/work/plplot-5.11.1 > sh: line 6: 95552 Trace/BPT trap: 5 ./test-drv-info mem > /home/pbulk/build/local/plplot/work/plplot-5.11.1/drivers/test_dyndrivers_dir/mem.driver_info > --- drivers/CMakeFiles/test_null_dyndriver.dir/all --- > *** [drivers/CMakeFiles/test_null_dyndriver.dir/all] Error code 2 > > bmake: stopped in /home/pbulk/build/local/plplot/work/plplot-5.11.1 > --- drivers/test_dyndrivers_dir/mem.driver_info --- > *** [drivers/test_dyndrivers_dir/mem.driver_info] Error code 133 > > bmake: stopped in /home/pbulk/build/local/plplot/work/plplot-5.11.1 > 1 error > > bmake: stopped in /home/pbulk/build/local/plplot/work/plplot-5.11.1 > --- drivers/CMakeFiles/test_mem_dyndriver.dir/all --- > *** [drivers/CMakeFiles/test_mem_dyndriver.dir/all] Error code 2 > > bmake: stopped in /home/pbulk/build/local/plplot/work/plplot-5.11.1 > Linking C executable pltek > A failure has been detected in another branch of the parallel make > > bmake: stopped in /home/pbulk/build/local/plplot/work/plplot-5.11.1 > --- utils/CMakeFiles/pltek.dir/all --- > *** [utils/CMakeFiles/pltek.dir/all] Error code 2 > > bmake: stopped in /home/pbulk/build/local/plplot/work/plplot-5.11.1 > A failure has been detected in another branch of the parallel make > > bmake: stopped in /home/pbulk/build/local/plplot/work/plplot-5.11.1 > --- drivers/CMakeFiles/xwin.dir/all --- > *** [drivers/CMakeFiles/xwin.dir/all] Error code 2 > > bmake: stopped in /home/pbulk/build/local/plplot/work/plplot-5.11.1 > 4 errors > > bmake: stopped in /home/pbulk/build/local/plplot/work/plplot-5.11.1 > *** [all] Error code 2 > > bmake: stopped in /home/pbulk/build/local/plplot/work/plplot-5.11.1 > 1 error > > bmake: stopped in /home/pbulk/build/local/plplot/work/plplot-5.11.1 > ~~ > __________________________ 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 __________________________ |
From: Jim D. <ji...@di...> - 2015-05-25 19:27:22
|
> On May 21, 2015, at 11:21 PM, Alan W. Irwin <ir...@be...> wrote: > > Hi Lewis: > > I am going to use a new subject line for this question for obvious > reasons. > > On 2015-05-21 17:51-0500 J. Lewis Muir wrote: > >> P.S. Even though the problem I reported appears to have been fixed, >> my build did not succeed, so I'm guessing I hit a new issue. The >> build failed as indicated below. Unfortunately, I don't have time to >> investigate, so this is really just an FYI. If you have people testing >> on OS X and things work for them, then it may be something related to my >> setup or building in pkgsrc. Or maybe you don't have people testing on >> OS X, and this is something that you'll eventually run into on OS X at >> some point before you make an official 5.11.1 release. This is on OS X >> Yosemite 10.10.3. > > The fact is we have suffered for a very long time from a lack of > testing on the OS X platform so some build bugs may be currently > present on that platform. However, if those exist it should be > absolutely straightforward to fix them once we know exactly what they > are. I do know some of our developers here have access to that > platform so I hope your report inspires them to do some testing on > that platform starting with using the VERBOSE=1 option for the make > command to find out exactly how our CMake-based build system > instructions translate to actual build commands on OS X. > > If some developer here with OS X build expertise is interested in > testing on that platform, my additional advice is to look at our > current override file which says (based on Mac OS X knowledge from > nearly a decade ago!) > I just built from the current repository on a Mac (OS X 10.10.3) using the steps from the documentation and did not get any errors. My build environment: CMake version 3.2.2 from MacPorts CC Apple LLVM version 6.1.0 (clang-602.0.53) (based on LLVM 3.6.0svn) Target: x86_64-apple-darwin14.3.0 Lewis, what is your build environment? |
From: J. L. M. <jl...@im...> - 2015-05-26 21:12:05
|
On 5/25/15 2:26 PM, Jim Dishaw wrote: > I just built from the current repository on a Mac (OS X 10.10.3) using > the steps from the documentation and did not get any errors. > > My build environment: > > CMake > version 3.2.2 from MacPorts > > CC > Apple LLVM version 6.1.0 (clang-602.0.53) (based on LLVM 3.6.0svn) > Target: x86_64-apple-darwin14.3.0 > > Lewis, what is your build environment? Hi, Jim. Thanks for looking into this. My build environment was pkgsrc-2015Q1 with CMake 3.2.2 backported from pkgsrc-current, and, I can't remember, I may have had to backport one other package too. The compiler was Xcode 6.3.1. If it builds fine for you, then forget it. I'll try again when pkgsrc-2015Q2 is released and report back if I still have a problem then. Regards, Lewis |