>  you should look at
> the archives to see what people decided.

from reading the archives and the response so far as well as the current 
documentation on the subject i conclude that the issue is still not clear 
and people make up their own stuff as they go along.

the following may be a formalisation of current (best) practices:

 - maintainers fall into three categories. herds, gentoo developers and
   non-gentoo proxy maintainers.

 -  a herd is defined in herds.xml. exception: no-herd. no-herd is limited
    to the situation where no suitable herd can be found.

 - a gentoo developer is defined in dev-rel/roll-call/userinfo.xml.

 - every ebuild has a herd.

 - a package can belong to more than one herd.

 - a package can be maintened by no or more parties.

 - a package with herd different from 'no-herd' is said to be maintained
   by the members of that herd.

 - all maintainers different from herds state their maintainership using
   the <maintainer> tag. the <maintainer> tag requires the <email> tag.

 - a package with herd 'no-herd' and no additional maintainers  is said to
   be unmaintained.

 - a <maintainer> of [EMAIL PROTECTED] indicates no maintainership and
   is only allowed as the sole maintainer. meaning a single <maintainer>
   tag and a single <herd>no-herd</herd> tag. infact it is semantically
   equivalent to a single <herd>no-herd</herd> tag and no additional
   maintainers.

 - a package that is proxy maintained needs an additional gentoo
   association in form of a maintaining herd or gentoo developer.
   [EMAIL PROTECTED] may be such an association, but only if the
   original one disappeared.

optionally one may want to include, although personally i prefer to state 
it explicitly:

 - a missing or empty <herd> tag is equivalent to <herd>no-herd</herd>

open questions:

 - what does it mean when the package is maintained by a herd and an
   additional maintainer? how does one determine the order in which to
   contact multiple maintainers?

 - are there situations in which we specify a <herd> other than no-herd
   but don't want its members to act as maintainers?

...

what are your current best practices regarding metadata?
regards
Thilo

Attachment: pgpA6IXbqq6cC.pgp
Description: PGP signature

Reply via email to