Re: Bug reporting [Was: Re: [PATCH] change gnats.html to fix web/8168]

["Joseph S. Myers" <jsm28 at cam dot ac dot uk>]

On Tue, 8 Oct 2002, Wolfgang Bangerth wrote:

> One of the unfortunate consequences would be that people start sending the 
> .i file of their original project, rather than trying to cut it down as 
> much as possible, if gcc would just set up everything for mailing upon 
> first hitting the ICE. I think, if I have to fill everything into a form 
> like on the web interface, including a description of what is going wrong, 
> I'm more tempted to isolate the problem. Given the shortage of people 
> fixing bugs, this seems like something one should encourage.

Much of cutting down should be automatable (with a little help from the
compiler to say where statements and declarations are in the source, so
that trials can be done automatically of removing them as far as possible
until no ICE can be triggered).  It hardly matters whether this automated
step is done it the user or developer end, except that doing it at the
user end may help with some cases where the full source is confidential.

> PS: Some time ago, Daniel Berlin had this nice copy of the GNATS database 
> converted to Bugzilla running. There was a long discussion, and the 
> general consensus seemed to be in favor of switching to Bugzilla. What 
> happened to that?

I thought it was waiting on a replacement gcc.gnu.org machine being set up
first as a test machine (with Bugzilla, and with all the Bugzilla scripts
and infrastructure in CVS) then later replacing the current system in
production.  The last test converted database I looked at seemed to be
working fine (apart from a few residual issues with the conversion of the
existing database - some being impossible to resolve in the conversion
because of GNATS discarding MIME headers necessary to properly decode
attachments to mails sent to GNATS in the past).

-- 
Joseph S. Myers
jsm28@cam.ac.uk