Hi Michael, 2015/12/31 Thu 2:10:14 UTC+9 Michael Soyka wrote: > On 12/30/2015 11:40 AM, Bram Moolenaar wrote: > > Ken Takata wrote: > > > >> 2015/12/30 Wed 22:46:09: UTC+9 Bram Moolenaar wrote: > >>> I wrote: > >>> > >>>> Ken Takata wrote: > >>>> > >>>>> 2015/12/29 Tue 21:59:54 UTC+9 Bram Moolenaar wrote: > >>>>>> Ken Takata wrote: > >>>>>> > >>>>>>> 2015/12/29 Tue 6:25:06 UTC+9 Bram Moolenaar wrote: > >>>>>>>> Christian Brabandt wrote: > >>>>>>>> > >>>>>>>>> On Mo, 28 Dez 2015, Bram Moolenaar wrote: > >>>>>>>>> > >>>>>>>>>> Patch 7.4.982 > >>>>>>>>>> Problem: Keeping the list of tests updated is a hassle. > >>>>>>>>>> Solution: Move the list to a separate file, so that it only > >>>>>>>>>> needs to be > >>>>>>>>>> udpated in one place. > >>>>>>>>>> Files: src/testdir/Make_amiga.mak, src/testdir/Make_dos.mak, > >>>>>>>>>> src/testdir/Make_ming.mak, src/testdir/Make_os2.mak, > >>>>>>>>>> src/testdir/Make_vms.mms, src/testdir/Makefile, > >>>>>>>>>> src/testdir/Make_all.mak > >>>>>>>>> Starting from this patch, this breaks appveyor: > >>>>>>>>> This built timed out after 60 minutes when trying to run test49: > >>>>>>>>> https://ci.appveyor.com/project/chrisbra/vim/build/204 > >>>>>>>>> After that patch, it fails when copying test49.in: > >>>>>>>>> https://ci.appveyor.com/project/chrisbra/vim/build/205 > >>>>>>>>> https://ci.appveyor.com/project/chrisbra/vim/build/206 > >>>>>>>>> > >>>>>>>>> Don't know, what the problem is and how to solve it. > >>>>>>>> I noticed. Somehow test49.in wasn't processed. It was not in the > >>>>>>>> list > >>>>>>>> of tests for MS-Windows before, so I removed it in patch 7.4.986. > >>>>>>>> > >>>>>>>> Now it appears it doesn't run any tests... > >>>>>>>> OK, I can reproduce it locally. > >>>>>>>> It's because appveyor doesn't specify a target, and the first > >>>>>>>> dependency > >>>>>>>> is in Make_all.mak: > >>>>>>>> > >>>>>>>> test49.out: test49.vim > >>>>>>>> > >>>>>>>> I'll specify a default target before including Make_all.mak > >>>>>>> After 7.4.988, still something strange happen. When running a single > >>>>>>> test > >>>>>>> as the following: > >>>>>>> > >>>>>>> nmake -f Make_dos.mak clean > >>>>>>> nmake -f Make_dos.mak VIMPROG=..\gvim test86.out > >>>>>>> > >>>>>>> Copy command fails like this: > >>>>>>> > >>>>>>>>> https://ci.appveyor.com/project/chrisbra/vim/build/206 > >>>>>>> Maybe this is caused by that testXX.out has multiple dependency rules. > >>>>>>> Attached patch fixes this. > >>>>>> That problem has already been fixed. We are now back to test 86 > >>>>>> failing. Before the diff was just one line, now it's a big difference. > >>>>>> On my system the test passes without problems. It might depend on the > >>>>>> Python version. > >>>>> No, it still occurs: > >>>>> https://ci.appveyor.com/project/k-takata/vim/build/37 > >>>>> I changed appveyor.yml to execute only test11: > >>>>> https://github.com/k-takata/vim/commit/d16eba3da82b8e7ba6dd672650eeb5b89225eab5 > >>>> Well, I don't know what your setup is, but for the main Vim repository > >>>> the builds pass now. > >> My setup is here: > >> https://github.com/k-takata/vim/commit/d16eba3da82b8e7ba6dd672650eeb5b89225eab5 > >> > >> I just commented out the following line: > >> > >> nmake -f Make_dos.mak VIMPROG=..\gvim > >> > >> and added the following line instead: > >> > >> nmake -f Make_dos.mak VIMPROG=..\gvim test11.out > >> > >> > >> This failure doesn't seem to happen when running all tests, but it happens > >> when running a single test. > >> This is very annoying for me now. > > Let me include your change. I suppose the order of the rules matters > > for the order in which the commands are executed. > > > >>>>> I wrote another patch. This is simpler than before, but I don't > >>>>> understand > >>>>> why this patch solves the problem: > >>>> Apparently the line you moved down must be after the rule that copies > >>>> the input files to dostmp. > >>>> > >>>> I wonder if we can get rid of this anyway, the tests should be able to > >>>> run with Unix file format. I wonder what files if we skip this step and > >>> s/files/fails/ > >>> > >>>> if we can fix that. > >> I'd like to try to run the tests with Unix file format when I have time. > > The DOS/MS-Windows files are currently distributed in dos fileformat. > > I'm considering dropping that in 7.5, so that the files are identical to > > what one gets from github. > > > Git has a configuration option "core.autocrlf" that enables/disables > linefeed-to-carriagereturnLinefeed and reverse conversions. > If the client-side sets this to true, this conversion is done > automatically by the client. The server-side has no control over what > the client does. The "ProGit" book discusses this chapter 8.1.
I don't think it works well. When I tried similar configuration on mercurial, it didn't work because some of the test input files contains some control codes including "\r". Some files were handled as binary files, and "\r" was converted to "\r\n". We also need to care about the mercurial mirror repository. > In view of this, it looks like Vim will have to continue to assume > responsibility for ensuring the appropriate line endings. Regards, Ken Takata -- -- 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.
