Re: Orange for Debian Med (Was: Help with my Debian package for Orange)
On Thu, Jun 30, 2011 at 02:41:51PM +0200, Mitar wrote:
> python-orange and orange-canvas.
OK, I adapted the tasks files accordingly and after next cron run
(tomorrow) this will be reflected on the tasks pages.
> > Well, there is other scientific software that is managing some kind
> > of release management. What exactly makes your software so specific
> > that this should not be possible? (just provoking :-))
>
> We do not really want to provide support for releases. We just fix it
> in trunk and users grab new version.
It does not finally matter whether you as upstream intend to support any
released code or not. At some point in time a version of your package
will end up in Debian stable. Debian as a distributor now becomes
responsible to care for this "release" regarding security etc (and it
would be great if you as upstream would provide at least help for the
security team when it is doing its job). If you ask me it looks way
more trustworthy if I see a release number attached to a package rather
some (random ??) SVN commit id.
Moreover I have a lot of scientists who insist on using a very specific
version of some piece of software because this (and only this? ... at
least these people are claiming so) can reproduce their scientific
results.
In Debian we also have uscan which is using debian/watch files to verify
that the last upstream version is packaged. So we somehow relay on
released and versioned tarballs to keep Debian up to date and you can
even see this reflected on the tasks pages (see the yellow "new version
available" buttons). You will not be able to profit from this system
if you intend to stick to use SVN.
>From my point of view it makes perfectly sense to tag a certain SVN
commit with a version number and finally it is your decision to say
that you do not support ancient versions.
Feel free to discuss this on Debian Science mailing list where such
discussions belong and where you can get more input into this (perhaps
with different opinions than mine).
Kind regards
Andreas.
--
http://fam-tille.de
Reply to: