As suggested on the ml lately, please consider making the custom 'no-herd' value deprecated in favor of specifying no <herd/> elements. As for backwards compatibility, right now all three PMs handle no <herd/> elements gracefully and only portage (half-)correctly supports 'no-herd' (i.e. treats it as a special value). If there's actually an app having problems with this, it probably has problems with multiple herds as well and we should fix it anyway. metadata.dtd allowed no <herd/> elements all the time. I think all that's necessary is updating the manuals then, and readding repoman check for no-herd value.
Current state: 1) handbook updated, 2) devmanual patch prepared, 3) skel.metadata.xml updated, 4) repoman enhancement request open. Portage never complained about lack of <herd/> elements so it's already clear to commit metadata.xml files without them.
Ok, the main tree should be fixed now and the development branch of Sunrise overlay as well.
01:46 <+CIA-88> antarus * gentoo/xml/htdocs/proj/en/devrel/undertakers/scripts/retire.py.txt: Update retire script to deal with no-herd properly now. 01:49 <+CIA-88> antarus * gentoo/xml/htdocs/proj/en/devrel/undertakers/scripts/retire.py.txt: Add missing equals, tested with python retire.py.txt -h FYI -A
BTW maybe we should consider replacing metadata.dtd with a more complete XML Schema or relax-ng?
(In reply to comment #4) > BTW maybe we should consider replacing metadata.dtd with a more complete XML > Schema or relax-ng? Please file a separate bug for that ;)
find gentoo-x86 -name "metadata.xml" -exec grep "no-herd" {} \; returns nothing so all packages are now fixed. Closing
