Re: Gitlab Migration

[Eli Zaretskii]

> From: Alexander Adolf <alexander.adolf@condition-alpha.com>
> Cc: emacs-devel@gnu.org, danflscr@gmail.com, Dmitry Gutov <dgutov@yandex.ru>
> Date: Thu, 02 Sep 2021 13:06:32 +0200
> 
> Thus, the magic words would seem to be "backwards compatibility", and
> "migration path".
> 
> "backwards compatibility" := Everything that works via email now, must
> equally work via email with the new system.

"Everything" is not a requirement, IMO.  It's nice to have, but we do
so much with email that requiring nothing to change is too much.
There should be a way to get notifications via email about the
important stuff, and there should be a way to participate in
discussing issues via email, including sending patches.

> "migration path" := It must be possible to try out one function on the
> new system, whilst still using all other functions via email.

Also not necessarily a requirement: it's okay if several functions
must be tried together.  Nice to have, certainly.

> Finally, a remark on issue/bug tracking. I have used a great many
> trackers, and debbugs and Bugzilla stand out from the crowd. By a huge
> margin. The differentiating feature is blocks/depends-on. Many issue
> trackers (including those in the various git hosting platforms) lack
> this feature altogether. Those trackers who have it, often provide a
> limited flavour, e.g. limiting it to one level (i.e. a bug that blocks
> another bug, can itself not be blocked by other bugs), or make it
> awkward to use (e.g. you have to create a new bug as a subtask of
> an existing one, and existing bugs cannot be made into subtasks any more
> once created).

Well, I agree that this is a good feature, but we don't use it too
much here.  So it is not high on my list of important features that
I'd consider as blocking the migration.

> Hence my plea: debbugs ought to remain part of whatever any new system
> might be.

Not if the new platform learns to support blocking, right?

Thanks.