[plt-dev] Re: launchpad instead of gnats?

On Dec 10, Sam TH wrote:
> On Thu, Dec 10, 2009 at 5:23 PM, Eli Barzilay <eli at barzilay.org> wrote:
> > On Dec 10, Sam TH wrote:
> >> On Thu, Dec 10, 2009 at 5:08 PM, Eli Barzilay <eli at barzilay.org> wrote:
> >> > On Dec 10, Sam TH wrote:
> >> >> On Thu, Dec 10, 2009 at 4:59 PM, Eli Barzilay <eli at barzilay.org> wrote:
> >> >> > On Dec 10, Eli Barzilay wrote:
> >> >> >> Questions to consider:
> >> >> >> [...]
> >> >> >
> >> >> > And more:
> >> >> >
> >> >> > * Is it possible to submit a bug in other ways than the web
> >> >> >  interface.  (An email interface will do.)
> >> >>
> >> >> Yes.  See here:
> >> >> https://help.launchpad.net/Bugs/EmailInterface
> >> >
> >> > I don't have time to read all of these in detail now, but this:
> >> >
> >> >  Launchpad only accepts email that is GPG signed.
> >> >
> >> > doesn't look good.
> >>
> >> Why not?  It's not that hard to have a key that DrScheme uses to
> >> sign generated bug reports.  Did you expect them to allow arbitrary
> >> people to email them without any authentication?
> >>
> >> They also have a (beta) web interface described here:
> >> https://help.launchpad.net/API
> >
> > * Is it common to distribute an application with a "private" key?
> 
> I don't know.  Ubuntu has an automated bug reporter called `apport'
> that you could look at.

You remember me saying that someone will need to do some work?  That's
part of it.  (And I thought that you were volunteering to do that.)


> > * Can you do the signing in Scheme code so it can run on Windows?
> 
> How could this not be possible?

Implementing encryption algorithms is difficult -- are you going to
write the scheme code to do that?  If not, you need to find some
libraries that makes it possible to do this on any architecture (and
everything that is involved).


> > * If all emails need to be signed, this means that everyone on the
> >  project need to use pgp?  (Nobody does, currently.)
> 
> For sending emails that are put on the bug tracker, yes.  And John
> uses GPG.

No, John uses the Apple thing -- and if it's incompatible with GPG in
some way that doesn't make launchpad happy, then he'll need to switch.


> > * What about email replies from users?  (I personally intensely
> >  dislike bug systems like SVN's that force me to use their web
> >  interface to say something.)
> 
> I think you're probably in the minority among users.

Huh?  There are many people who reply with no problem to bug emails.
Also, I know about the usual set of complaints against gnats, and
communicating over emails was never on the list.  In fact, I suspect
that people will still do a lot of such communication in emails, which
means that information will not appear as part of the bug unless it
(the launchpad bug system) can be used with such reply emails.


> However, the larger question is, do we want to have a modern bug
> tracker that will make it easier for people to contribute to the
> project and for developers to manage their bugs, or do we want to
> have core members of the group continue to spend their time on web
> application administration in Perl?

Just to put things in perspective, the current system requires very
little administration work, and switching to launchpad sounds like a
sizeable effort.  If I were you (aka the person who volunteers to do
all of this), then I'd be extremely careful in estimating the kind of
work involved.


> If it's the former, we should consider our options, and decide among
> them, rather than just listing potential problems.

As the person who will do all of this work, listing potential problems
is something that you should appreciate -- since these are all things
that you'll need to deal with.

(For example, if launchpad doesn't do nag emails, and if "management"
considers them important enough as it has in the past, then you will
need to write the code that somehow extracts that information and
produces these emails.  Pointing at the lack of that functionality
should therefore be helpful in your decision on entering such a
commitment; something that is much better to do now rather than when
it's too late.)

-- 
          ((lambda (x) (x x)) (lambda (x) (x x)))          Eli Barzilay:
                    http://barzilay.org/                   Maze is Life!