Re: gnats

[Yngve Svendsen]

I've been bitten by the "bad responsible file" problem a few times myself. 
As regards version 4 of GNATS, I have some information that might be of 
comfort. Although GNATS still doesn't do much syntax checking and has no 
facility for reporting misconfiguration, the parsing of the config files is 
more robust, so you won't see this particular problem in version 4.

I think the easiest way to solve this type of problem is to create a 
utility which checks the GNATS config files and reports any problems it 
finds. Kind of like the -t option to Apache httpd.

But I fear that we have few resources for implementing this. Any volunteers 
out there?

Yngve Svendsen
Gnatsweb maintainer


At 17:46 01.08.2001 -0600, Tom Tromey wrote:
> I had a frustrating Gnats adventure today.
> Hopefully someone can learn from my experience.
>
> Automake PRs stopped arriving.
> `queue-pr --run' silently failed -- deleting the new PR.
> (This took a while to discover)
>
> So then I tried running file-pr.  It SEGVd.
> (This also took a while to discover)
>
> Finally I ended up running gdb on the file-pr that still has debug
> symbols in it.  Turns out that the automake `responsible' file had an
> error in it.
>
> I'd really, really prefer it if
> (1) gnats sent email somewhere saying "I'm misconfigured"
>     if an error is found
> (2) queue-pr complained if file-pr failed
> (3) queue-pr did *not* delete a pending PR if file-pr failed
> (4) more error checking were done on input files.
>     In my case, init_responsibles() SEGVd because I had an entry in
>     `responsible' with just two fields.
>
> Unfortunately I don't have time to work on these things.
>
> For Gnats folks, we're using 3.113 on sources.