This is the mail archive of the mailing list for the GCC project.

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]
Other format: [Raw text]

Re: [Patch] Remove gccbug from bugreport.texi

On Tuesday, May 13, 2003, at 12:25 PM, Wolfgang Bangerth wrote:

Joseph, I don't think your answer solves anything. First:

I don't either.


There are large parts of the world where the normal (or only available)
form of network access is dialup, charged by the minute, so WWW bug
reporting is significantly more expensive for those users than composing a
message with gccbug offline.

I think that's inappropriate, too. I was living with a 14k modem until
last year. It was fast enough to sometimes read the newspaper. If I can
think of a lean web page, then it's certainly the gnats submission form.
Given that the typical attachment for a bug is the range of >100k, I
seriously doubt that the overhead of http is significant there. After all,
the attachment also has to be sent when using mail.

The number of bugs filed by email has been decreasing over time.

Dec 2002:44 out of 355 (12.39%)
Jan 2003:77 out of 424 (18.16%)
Feb 2003:34 out of 368 (9.23%)
Mar 2003:55 out of 408 (13.48%)
Apr 2003:23 out of 306 (7.51%)

The numbers are a bit misleading:
In particular, Wolfgang accounts for over 15% of the email reports in January.
Also, all but 6 bug reports (total) are from people who also submit reports by web. is the only person i can find who has a significant number of reports, and *seems* to prefer submitting reports by email:
82 X-Send-Pr-Version: 3.113
1 X-Send-Pr-Version: gnatsweb-2.8.1 (1.28)
9 X-Send-Pr-Version: gnatsweb-2.9.3 (

If we were to adopt this line of reasoning, we would certainly have to
reject the move to bugzilla, since it has this big picture with the ant on
its main page.
It's an 8k gray scale jpeg. It takes roughly 4 seconds on a 14.4k modem.

Bug reports submitted through GNATSweb are often missing something as well
- the version number, or the attachment (through the submit defective bug
report / error comes back but for security the attachment needs to be made
manually on the form again route?).

That's a no-argument too. If both are bad, that's no reason to keep both,
right? Though the gnats bug will be fixed when moving to bugzilla, the
problems with gccbug will remain.

I'd like to stress again that I don't want to kill gccbug at all. I'm just
removing it from the list of recommended ways of submitting PRs. I'm fine
with people using stone age tools, but if it creates problems for us we
don't have to encourage it.

Which is why i support processing incoming gccbug reports, but wouldn't cry if it went away.


----------------------------------------------------------------------- --
Wolfgang Bangerth email:

Index Nav: [Date Index] [Subject Index] [Author Index] [Thread Index]
Message Nav: [Date Prev] [Date Next] [Thread Prev] [Thread Next]