repoman has problems when development is done for a different plattform (i.e. x86 on a sparc machine) Under some situations it would be really good to be able to commit even when the current directoy contains some problems for other archs. Some suggested options for repoman would be: --fake-arch or/and Setting the envs ARCH="foo" and/or ACCEPT_KEYWORDS="~foo" --ignore-arch did not work (Maybe keyword based if "~arch" and "arch" cause problems too) --force-commit It's absolutly clear that this is dangerous and should never be default in any script etc. But under some circumstances it can be really annying when you cannot commit. Also, sometimes you cannot fix everything and it's better to do your part and then hand it over to the person who is able to fix the rest. (like when you're only adding a stable keyword for an arch and the ebuild has other problems already not caused by this) Reproducible: Always Steps to Reproduce: 1. 2. 3. Actual Results: <no info> Expected Results: <no info> <no info>
Even kernel wont commit now because the optional doc useflag needs stuff which is not "sparc" keyworded now. Noone ever seemed to need that USE flag on a sparc machine yet. And that transfig would need tonnes of other libs incl. freetype etc. RepoMan scours the neighborhood... DEPEND.bad 2 sys-kernel/sparc-sources/sparc-sources-2.4.20-r8.ebuild: ['media-gfx/transfig'] sys-kernel/sparc-sources/sparc-sources-2.4.21-r1.ebuild: ['media-gfx/transfig'] RDEPEND.bad 2 sys-kernel/sparc-sources/sparc-sources-2.4.20-r8.ebuild: ['media-gfx/transfig'] sys-kernel/sparc-sources/sparc-sources-2.4.21-r1.ebuild: ['media-gfx/transfig'] repoman should really act like it did before and NOT enable every single keyword which enables tonnes of optional stuff mostly not available on sparc (maybe other non-x86 archs as well) Quick workaround would still be a --force-commit. Ugly but without we're almost unable to work on sparc with repoman.
re-assigning to puggy as he is the main repoman guy now.
Nakano is working on this. Describe in more detail so we know the issues you're running up against. Optimally, you'd get everything up to speed on sparc and this wouldn't be as much of a problem. profiles/*/use.mask exists to disable use flags that are not valid on your systems. Current. cvs/2.0.50 has a patch to help out with this, but we're still working on things.
recent repoman can check in each arches.
