Re: Offer to maintain GNU's Debbugs instance (and modernize it)

[Bob Proulx]

Felix Lechner wrote:
> I'd like to separate the GNU instance from the upstream project at
> Debian. My goal is to gradually reimplement Debbugs for the GNU project
> in GNU Guile. It would likely be a dynamic web application. Is that an
> acceptable path forward for everyone?

First let me say, wow, that would be an ambitious project!  Let me
cheer you on for it.  If I can help move that forward let me know what
I can do and I will be glad to help.

I don't think trying to replace the current GNU debbugs BTS in place
with a new web UI and with a rewrite in guile *in place* is the best
way forward though.  Have seen that happen many times with many
projects over the years and it has never worked out well for anyone.
Let's insert right here the long list of projects that have tried
doing a full in place teardown and rewrite of a live working system
with the result being quite problematic without actually detouring
through that discussion at the moment.  It's a long list.  It's
problematic.

Given what you have said already about wishing to make such a large
diversion from the Debian BTS then so far it sounds like more a new
project with a new vision rather than an upgrade of the existing
system.  As you yourself said in a later message you wish to give it a
completely overhaul.  That sounds to me like a new project.  Working
with the legacy of the existing users would feel like an obstacle.
Making it better to start new from the beginning.  I don't see the
existence of the current BTS to be a reason one can't create a new
different bug tracking system.  Just like the existence of cvs did not
stop the creation of svn and then the creation of git.  Just like the
existence of the Savannah Tracker did not stop the creation of the GNU
BTS.  It does not need to be a mutation of the existing system.

Instead I suggest to just start designing the system you want.  Work
steadily toward the point where it can self-host itself.  Then use it
to track bugs in itself.  And then proceed from there.  Honestly if
the result is good then you will have projects wanting to use it
instead of other alternatives and they will come to you asking to join
up and use it for their projects.  That's the best way!

Bob