[darcs-users] An idea of a bug reporting system

['Andrew Pimlott']

On Mon, Aug 16, 2004 at 10:53:18AM -0700, Dustin Sallings wrote:
> 
> On Aug 13, 2004, at 16:12, 'Andrew Pimlott' wrote:
> 
> >Another problem:  Usually, when a bug is opened, you don't know which
> >patch caused it, so you can't depend on it.  Only after someone figure
> >out which patch is responsible can he update the bug with a bug patch
> >that depends on the faulty code patch.  So if you have a repository
> >without that code patch and you pull all the bugs, you will get the
> >bug opening, but not the subsequent updates.
> 
> 	I've seen this argument a couple of times, but I don't quite get it. 
> I've never used a bug tracking system where anyone's actually tracked 
> back the exact checkin to associate as the ``cause'' of a new bug.  

A good point.  Here's why I made that argument:  It seems to me that the
"killer feature" that people are looking for by putting the bug database
in the repository is that the bugs are (in some sense) in sync with the
code.  I am trying to point out that this may not be as feasible as it
sounds.  You are correct that we shouldn't discard the idea just because
some "blue sky" feature is not totally achievable, but in that case ...

> I agree it'd be really neat to have your bug system tell you the 
> exact line of code that caused any given bug, but it seems very impractical 
> to avoid making a distributed bug system where bugs are associated with 
> their respective code just because it can't do this.  It seems like 
> we're trying to say we can't do this because it doesn't solve problems 
> that other systems don't solve anyway.
> 
> The only potential problem I see is that a ``released'' tree 
> wouldn't necessarily know about every bug that was reported against it 
> after the fact.  In that case, you'd just need to maintain a branch from 
> that tree and merge in bug reports when you find that they apply to that 
> branch.  It's still manual, and not all that great, but the problem 
> doesn't seem like an incredibly common one to me in the first place.  
> At least, not enough to outweigh the benefits of a distributed bug 
> tracking system for distributed projects.

... what advantage does this have over simply putting the bug DB in a
separate darcs repository?  I think that ultimately, this will be more
flexible, and you don't have to worry about merging the bug DB
separately from the code (which darcs doesn't support well anyway).

Andrew