On 14 November 2016 at 20:54, Peter Hutterer <peter.hutte...@who-t.net> wrote: > On Mon, Nov 14, 2016 at 03:39:18PM +0000, Emil Velikov wrote: >> On 13 November 2016 at 22:24, Peter Hutterer <peter.hutte...@who-t.net> >> wrote: >> > On Fri, Nov 11, 2016 at 03:45:29PM +0000, Emil Velikov wrote: >> >> From: Emil Velikov <emil.veli...@collabora.com> > [...] >> >> + return 1 >> >> fi >> >> - build_dir=`dirname $status_file` >> >> + >> >> cd $build_dir >> >> if [ $? -ne 0 ]; then >> >> echo "Error: failed to cd to $MODULE_RPATH/$build_dir." >> >> @@ -377,6 +373,15 @@ process_module() { >> >> return 1 >> >> fi >> >> >> >> + # Using ../ here feels a bit nasty, yet $top_src is an absolute >> >> path. Thus >> >> + # it will get propagated in the generated sources, which we do not >> >> want. >> >> + ../configure >/dev/null >> >> + if [ $? -ne 0 ]; then >> >> + echo "Error: failed to configure module." >> >> + cd $top_src >> > >> > I'd really like to see more pushd/popd, but that's unrelated to this patch. >> > Looks good, but I think you should look at the effort to do a full local >> > clone, it may only be a few lines and it should remove quite a bit of the >> > error checking. >> > >> Yes I'm thinking/working towards having things in a cleaner state. At >> the same time this patch would cause a noticeable change in workflow, >> so I've intentionally opted out of making things too intrusive. >> >> Any objections against git worktree and/or moving towards it as a >> follow-up change ? >> Or you'd thinking this patch should be "it's all or nothing" kind of change ? > > simply said, I don't know enough about worktrees to have an opinion here, > sorry. for the mere tarball generation and distcheck run you can use a > shallow clone though, can't you? Just something to keep in mind, if the > worktree works as a clean checkout without detritus, I'll be happy with tha > too. > In a line: worktree like a fresh clone, with the only difference being that .git is a file pointing to the actual git repo. Here is an example and more info [1].
Can we have all that [shallow/full clone, worktree, other] as follow-up, please ? Or you really don't like the git clean suggestion ? The current patch/workflow has been tested for over a dozen times and I feel a bit reluctant in pushing for something that's barely tested :-\ Thanks Emil [1] https://git-scm.com/docs/git-worktree#_examples _______________________________________________ xorg-devel@lists.x.org: X.Org development Archives: http://lists.x.org/archives/xorg-devel Info: https://lists.x.org/mailman/listinfo/xorg-devel