05 апр. 2015 г. 1:14 пользователь "Adam Wolk" <adam.w...@koparo.com> написал: > > Hi ports@ > > I was asked by upstream to pull up all patches up to: > - 1ea5df13f908495df4ccccad9d634d997f6fd4c9 > while the tagged revision is > - c1819f1184ae18a76ab455caa6fcc713c9c6bf3d > > This results in a 2k lines patch: > $ git diff 1ea5df1..c1819f1 | wc -l > 2289 > > minus the one bumping the version numbers. > > Spanning over 47 files: > $ git diff 1ea5df1..c1819f1 | grep '+++' | wc -l > 47 > > The porting manual suggests: > - "Only patch one source file per patch file, please." > > You probably already guessed that those commits are fairly > small but do span over multiple files per commit. > > How should I approach this? > > - generate one jumbo patch for bumping the revision > - generate patch per file loosing context > - generate a patch per commit reflecting git history > - package a potentially broken version (upstream states > issues as critical) > - any other ideas? > > I would really appreciate feedback here.
If I understood you correctly, the cumulative patch you're talking about gonna be removed after next version update (which should happen soon), right? I think you could go with a single patch then. Remember that this would make your life harder if you'll ever try to run make update-patches. ... Or you can just take a snapshot of "fixed" commit directly from Github using GH_* variables. :) Just keep the version in PKGNAME the same, bumping REVISION only. -- Vadim Zhukov
