Christian Brabandt wrote: > On Do, 18 Feb 2016, Bram Moolenaar wrote: > > > Christian Brabandt wrote: > > > On Mi, 17 Feb 2016, Bram Moolenaar wrote: > > > > > > > Thanks. I think we should also build a console version. Since testing > > > > appears to the take bulk of the time we can run the tests once with > > > > HUGE-console features. So perhaps we can build TINY-console with MSVC > > > > and HUGE-console with MingW and then run the tests with the MingW > > > > binary? That way we cover more rows & columns in the build & test > > > > matrix. > > > > > > Well, a console version is build for MSVC, since that appears to build > > > faster than MingW. Testing happens with the gvim.exe build from MSVC, > > > because first of all, the test I tried with MingW did stall on test47 > > > and second of all, last time I checked with the nightly windows builds, > > > the gui version test was a lot faster than the console test. > > > > Running the tests in the console is different from running them in the > > GUI, I think we should run both. > > Okay, we can do this. Will take longer however.
I'm glad you are taking the time to make this work. Pity it's not that easy. I see the current version still fails in some ways it would not fail when run locally. > > If a test stalls then we may have to skip it. We already skip several > > tests for MS-Windows, either they are omitted in the Makefile or the > > test bails out early. > > This only happened for the tiny version. Interestingly, it does not > stall, when I run it locally. Hm, let's see, if WINVER=0x501 as Ken > suggested fixes this. Also the tiny mingw version stalled at compiling > here: > https://ci.appveyor.com/project/chrisbra/vim-ch0ci/build/48/job/2ayvxobvq80uc551 Also see some warnings there. Let me know if you see more (so long as it's predictable how to fix them :-). It appears to get stuck in "windres". Perhaps it's showing a dialog and wants you to click OK. Hmm, the "Leaving directory" indicates that windres finished, thus it must be something else. The tiny build should probably not build GvimExt. And the console version should not build it either. > ,---- > | mingw32-make.exe -C GvimExt -f Make_ming.mak CROSS=no CROSS_COMPILE= > CXX='g++' STATIC_STDCPLUS=no > | mingw32-make.exe[1]: Entering directory 'C:/projects/vim-ch0ci/src/GvimExt' > | g++ -O2 -DFEAT_GETTEXT -c gvimext.cpp -o gvimext.o > | windres --preprocessor="g++ -E -xc" -DRC_INVOKED --input-format=rc > --output-format=coff -DMING gvimext_ming.rc -o gvimext.res > | g++ -shared -O2 -s -o gvimext.dll \ > | -Wl,--enable-auto-image-base \ > | -Wl,--enable-auto-import \ > | -Wl,--whole-archive \ > | gvimext.o gvimext.res gvimext_ming.def \ > | -Wl,--no-whole-archive \ > | -luuid -lgdi32 > | mingw32-make.exe[1]: Leaving directory 'C:/projects/vim-ch0ci/src/GvimExt' > `---- > Don't know, why this is happening. > > (a little bit later...) > Hm, no does not help. The tiny version still stalls there: > https://ci.appveyor.com/project/chrisbra/vim-ch0ci/build/job/fb8s0g2umwldgrue > > Also, latest mingw version still shows some warnings: > https://ci.appveyor.com/project/chrisbra/vim-ch0ci/build/55/job/nbyergs228easqat#L127 > > And since we are at it, the console version fails: > https://ci.appveyor.com/project/chrisbra/vim-ch0ci/build/55/job/nbyergs228easqat#L2538 > https://ci.appveyor.com/project/chrisbra/vim-ch0ci/build/55/job/nbyergs228easqat#L4062 > https://ci.appveyor.com/project/chrisbra/vim-ch0ci/build/55/job/nbyergs228easqat#L4157 > https://ci.appveyor.com/project/chrisbra/vim-ch0ci/build/55/job/nbyergs228easqat#L4662 > https://ci.appveyor.com/project/chrisbra/vim-ch0ci/build/55/job/nbyergs228easqat#L4666 > https://ci.appveyor.com/project/chrisbra/vim-ch0ci/build/55/job/nbyergs228easqat#L4670 > > > > I'm surprised the console tests are slower, the GUI has to open a new > > window, create menus, etc. Perhaps it's because console I/O is being > > slow. > > Yes, it is veeery slow. One Feature Set test (building gvim and vim > versions for MingW and MSVC and Testing MingW vim.exe and MSVC gvim.exe) > now takes about 20 Minutes. Hm, perhaps, we should add some time stamps, > before the test starts. > > > > > > > Perhaps the build can be made faster by not using -O3. -O2 should be > > > > sufficient to get useful warnings. Even -O1 should be enough and quite > > > > a bit faster. > > > > > > I did change that. Now MingW builds with -O2 which is indeed quite > > > faster and I believe it did generate some more warnings. > > > https://ci.appveyor.com/project/chrisbra/vim-ch0ci/build/47 > > > > Hmm, I would expect fewer warnings. Looks like they are easy to fix > > though. > > > > > Updated patch attached. > > > > This builds four versions and tests two? > > Building gui Mingw, gui MSVC and console MSVC each for huge and normal > feature set. The huge feature set also enables CHANNEL and PYTHON > version. > > > I do think we also want MingW-console and MSVC-GUI. > > Okay. Updated patch attached. -- "To whoever finds this note - I have been imprisoned by my father who wishes me to marry against my will. Please please please please come and rescue me. I am in the tall tower of Swamp Castle." SIR LAUNCELOT's eyes light up with holy inspiration. "Monty Python and the Holy Grail" PYTHON (MONTY) PICTURES LTD /// Bram Moolenaar -- [email protected] -- http://www.Moolenaar.net \\\ /// sponsor Vim, vote for features -- http://www.Vim.org/sponsor/ \\\ \\\ an exciting new programming language -- http://www.Zimbu.org /// \\\ help me help AIDS victims -- http://ICCF-Holland.org /// -- -- You received this message from the "vim_dev" maillist. Do not top-post! Type your reply below the text you are replying to. For more information, visit http://www.vim.org/maillist.php --- You received this message because you are subscribed to the Google Groups "vim_dev" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. For more options, visit https://groups.google.com/d/optout.
