Re: no more package moratorium?

["Robert Collins" <robert dot collins at itdomain dot com dot au>]

----- Original Message -----
From: "Corinna Vinschen" <cygwin@cygwin.com>
>
> > It does raise an interesting point: who, when, and how, do new
packages
> > get approved?
>
> That's a problem when getting lots of new packages.  The forum for
> discussion and the approval process is cygwin-apps.  However, it's
> not the forum to send loads of tar archives so we will have to find
> some standarized way as, just as an example:

Tarballs - package quality - are orthogonal to the discussion I was
raising. I've trimmed those aspects out in replying.

> - Potential contributor announces on cygwin-apps that s/he wants
>   to contribute package `foo' with a short description what the
>   package does.

I agree. They must also *At this point* agree to maintain the package do
upgrades feed patches to the vendor etc, and that they will announce
publicly if they decide to stop maintaining the package with as much
warning as possible. Packages with no maintainers are pulled after 3
months.

> - cygwin-developers discusses if the package should become part of
>   the distro and chooses a person from cygwin-developers as approver.

Nope. I don't think this is appropriate. cygwin-developers is for
developers of cygwin1.dll. Last I heard, Linus has no input into what
Redhat put into the (say) the RawHide distro, so why should the
cygwin1.dll developers care what goes into 'cygwin the net
distribution'.

I think we should either get a consensus from all the package
maintainers, or perhaps, wait 3 days for objections. If no objections,
then the package is allowed in. If there are objections, discuss until
resolved. To prevent deadlock, a single individual objecting will not
cause a package to be rejected, the objections must be agreed with by
other package maintainers.

Some sort of voting thing might be nice (mentioning to show I've thought
about it) but for now it seems too hard for too little benefit. I do
like the idea of a sponsor, so

once a package is decided to be allowed in, if its the first package
from the maintainer (ie a new maintainer) then an existing maintainer
must sponsor the package, and vet package quality -
textmode/patches/postinstall scripts etc.

> - When the approver thinks the package is ok,  the contributor
>   is (obligatory!) asked if s/he's willing to maintain the package
>   in future and if s/he's willing to announce officially when
>   s/he's not anymore willing to maintain the package.

Good points. modified slightly

> - When the contributor/maintainer announces to drop maintainership,
>   we will ask for another person willing to maintain the package
>   further.  If we don't find another person within, say, three months,
>   the package will be removed from the distro.

As you can see above, this does not cover getting the tarball into the
net distro: as I said, thats orthogonal.

I think the process for that part should be something like

sponsor (for new maintainers) or maintainer (2nd package or new version
of existing) places the packages files at a URL.
They tell someone from <list of maintainers with write access>.
<someone> uploads to cygwin.com.

If there is _any_ doubt about the package quality, upload it as
experimental. Wait 3 weeks, and if there are no bugs reported, then edit
setup.hint to make that new versiom current.

Thoughts on this?

Rob


--
Unsubscribe info:      http://cygwin.com/ml/#unsubscribe-simple
Bug reporting:         http://cygwin.com/bugs.html
Documentation:         http://cygwin.com/docs.html
FAQ:                   http://cygwin.com/faq/