Bug to help me track other mercurial.eclass related issues. I'm working on heavy modifications of mercurial.eclass, expect patches attached to this bug soon.
Created attachment 182775 [details, diff] mercurial.eclass.patch mercurial.eclass path, changes: - fixed (hopefully) existing bugs - eclass documentation - general cleanup
Against which version do you patch? (the patch failed for me in the tree as well as in my local version)
From the patch itself: diff -u -r1.4 mercurial.eclass Could you post output of patch with mercurial.eclass.rej file ?
I don't work on the cvs but on my rsync tree mirror...
And the sync I just did fixed it.
I've been testing this patch, but it has teh disadvantage of not offering a Mercurial repository for packages. The packages might require being in a repository for building, though. For example the Mercurial setup.py checks the id of the repository to set its version dynamically. What is your reason for using archive instead of clone?
Created attachment 184238 [details, diff] mercurial.eclass.patch This uses clone instead of export.
With this patch, dev-util/mercurial-9999 builds again. Many thanks!
(In reply to comment #8) > With this patch, dev-util/mercurial-9999 builds again. > > Many thanks! Commited.
I just spotted one minor point: In line 103 of the eclass the error message should be "hg clone failed" instead of "hg archive failed". 103: "${WORKDIR}/${module}" || die "hg archive failed" Also you don't need "rm -rf" to remove a Mercurial repository. A simple "rm -r" suffices (I don't know if that's relevant, though).
(In reply to comment #10) > I just spotted one minor point: > > In line 103 of the eclass the error message should be "hg clone failed" instead > of "hg archive failed". > > 103: "${WORKDIR}/${module}" || die "hg archive failed" Thanks, fixed.