On 2022/02/08 20:32, Andrew Krasavin wrote: > On Tue, Feb 08, 2022 at 12:31:39PM +0000, Stuart Henderson wrote: > > > > > +MASTER_SITES0 = https://github.com/google/googletest/commit/ > > > +# Fix gtest-help-test failure on OpenBSD > > > +# https://github.com/google/googletest/pull/3751 > > > +PATCHFILES = fix-gtest-help-test-{}631f4f99.patch:0 > > > > it just touches a couple of lines in one file. why not use a > > normal ports patch? > > Thanks for the feedback! > > Personally, I thought that using PATCHFILES was always preferable to > adding normal ports patches. Apparently this is not the case, and in > cases where the changes are minor and affect a single file, 'normal ports > patches' in the 'patches' directory should be preferred?
I would only use them if they were really complex to manage in the ports tree. If a file needs further changes it's simple to do it where patches are in the ports tree, and a pain if they're fetched from another source. In particular with the ones from github, they're subject to change if the software producing them changes, or if the shortform commit hash is lengthened (I've seen that before in ports). > In any case, my pull-request is now merged into main branch, so, you > can just delete cited code and set GH_COMMIT to > '06519cedc3159de8b36a504766ad6b7966555f10'. > > (I will be able to send you the patch when I will be at my > openbsd-machine, right now it is, alas, not so) Thanks :)
