Re: GNOME 1.4 extra apps coordinators

[Malcolm Tredinnick <malcolm commsecure com au>]

On Sun, Oct 22, 2000 at 03:46:20PM -0500, Dan Mueth wrote:
> I am still for having gnome-xxx-extras packages. I promise not to scream
> and shout if everybody disagrees, but I will try to explain why.
> 
> The main reason is for the user.  Right now the user has a hard time
> finding additional good applications for their system.  The software map
> helps some, but it is somewhat tricky to navigate.  A search feature would
> help here.  Once one does find an application they want, the quality is
> completely random.  The application could be long abandoned, only
> available in certain packages (eg. only tgz), and while I haven't done a
> careful survey, my impression is that there is little translation or
> even documentation of these packages.  Having all of the applications
> seperate means the user must find, download, install, and test each
> application and hope for the best.  Having a single package of tested and
> supported applications would help a lot here.

One of the big arguments against is for the user, as well,
unfortunately. A gnome-xxx-extras package is likely to be huge. Many
users still have to download things over modems and/or pay for things
above a given download limit (I am in the former category at home, for
example).

> Now having the extra apps will help, since a certain level of QA will be
> done on these packages before giving them the "stamp of approval" of the
> extra apps coordinators.  Will the extra apps coordinators make sure each
> app is available in tgz, rpm, and deb packages?  This would be very nice,
> but would take some work.  We also would like to have documentation and
> translations for each package.

The creation of packaes for each distribution _is_ an argument against
the "many small packages" plan and I don't think the burden should rest
with the extra apps coordinators. I'm not sure what the solution is
here; it is difficult for developers to create packages for each
distribution, just because they won't have the facilities to test it.
Maybe we need to encourage a "packaging group" who do a similar job to the
i18n group (who run around translating everything in site).


> This leads me to the other group having gnome-xxx-extras packages serves -
> the developers.  If we put these packages in CVS, then everybody can work
> together on them.  Hackers can go in and fix bugs.  Doc writers can go in
> and add docs.  Translators can go in and add translations.  Packagers can
> make debs or RPMS. Et cetera.

A lot of the things are in gnome CVS or Source Forge CVS anyway, so I
don't quite follow this argument. And, aside from reponse time problems,
once you have checked out a CVS tree, you never have to worry about
where you got it from again; so whether it's from Gnome or Source Forge
or "Joe's Fly By Night CVS", it shouldn't really matter.

An argument against putting everything into one big package is bug
reporting -- people will report a bug in gnome-xxx-extras, rather than
in application "foo". So it will be more difficult for the maintainer of
foo to track bugs against their app.

> While developers have the best intentions, experience shows that in
> many cases they are typically so busy that they are slow and
> unreliable at integrating documentation, translations, patches, etc.
> which people send them by email.  CVS works *much* better in my
> experience.  Overall I think the workload is reduced and the quality
> is improved if these apps reside in GNOME CVS where everybody can work
> together on them.

Many maintainers do not want people just committing willy nilly to their
package at the moment -- they would rather have the patches run past
them once just to check that things don't get broken due to a
misunderstanding or anything.

Although Dan is probably in a better position than myself to know this,
I was under the impression that the committing of documentation,
translations, et al, for those packages that were in CVS worked quite
well at the moment. The translation stuff was incorporated pretty much
automatically and it was up to the documenters to let the maintainer
know when documentation was comitted and ready for integration. Or am I
mistaken?

Cheers,
Malcolm

-- 
Malcolm Tredinnick            email: malcolm commsecure com au
CommSecure Pty Ltd