Build fails due to possibly a bug in portageq: !!! In portageq at line 1647 !!! portageq emulation for get_repos not implemented Reproducible: Always Steps to Reproduce: 1. use paludis 2. cave resolve -1 app-portage/eclass-manpages 3. failed build Actual Results: See attached build log. Expected Results: Successful build. Is it kosher to use portageq from inside of an ebuild? I may be reading the output incorrectly as well, just wondering.
Created attachment 422308 [details] build output
> Is it kosher to use portageq from inside of an ebuild? I may be reading the output incorrectly as well, just wondering. Previously, it referred to PORTDIR_OVERLAY, which is also portage-specific. That variable has been deprecated for several months. eclass-manpages has never really been PMS-compliant. I believe there no supported way of generating a list of enabled repositories without relying on package manager implementation details.
not a whole lot to be done here. tools can't rely on PORTAGE_OVERLAY anymore as everything is migrating to `portageq get_repos`. i guess ask the paludis guys to implement that.
Why does the ebuild even scan overlays? That isn't the behaviour I would expect. If I add an overlay to layman because I need one specific package from it, then I don't want it to influence other packages by some side effects. In other words, eclass-manpages in the main tree should only install man pages for eclasses in the main tree. Anything overlay specific should be installed by a dedicated package in that overlay.
The problem now applies to keyword-masked live ebuild only. Do you think it'd be reasonable to have a USE flag to control this behavior?
(In reply to Michał Górny from comment #5) > The problem now applies to keyword-masked live ebuild only. Do you think > it'd be reasonable to have a USE flag to control this behavior? Simply drop the live ebuild. Its days are counted anyway, since ECLASSDIR is gone in EAPI 7. (Plus, it currently fails to build.)
Hmm, or I guess I could make it a git live ebuild, with new git-r3 features.
The bug has been closed via the following commit(s): https://gitweb.gentoo.org/repo/gentoo.git/commit/?id=0dded1c44940fa37150be78ea1ae4cc0cf5bbf95 commit 0dded1c44940fa37150be78ea1ae4cc0cf5bbf95 Author: Michał Górny <mgorny@gentoo.org> AuthorDate: 2019-03-14 12:35:21 +0000 Commit: Michał Górny <mgorny@gentoo.org> CommitDate: 2019-03-14 12:35:21 +0000 app-doc/eclass-manpages: Switch the live ebuild to use git Closes: https://bugs.gentoo.org/571284 Signed-off-by: Michał Górny <mgorny@gentoo.org> .../eclass-manpages-99999999.ebuild | 35 +++++++--------------- 1 file changed, 10 insertions(+), 25 deletions(-)