From: mdw@theory.TC.Cornell.EDU (Matt Welsh) Subject: Re: RFD: Bug Forms for Linux Date: 27 Apr 1993 01:51:38 GMT
In article <gilC6426o.JrE@netcom.com> gil@netcom.com (Gilbert Nardo) writes:
><brief> This is a request for discussion on a standard linux bug report form.
>
><reason> A large percentage of the traffic in comp.os.linux consists of
> unstructured bug reports. There exists a problem whereby a person
> interested in solving a bug does not receive minimal contextual
> information in the report, and usually posts or mails follow-up
> questions. Additionally, time is unnecessarily spent deciphering
> relevant information from noise for each particular report.
You may wish to speak with Fred Van Kempen (waltje@uwalt.nl.mugnet.org)
regarding this. He is planning on providing a bug-reporting mechanism to
reduce the traffic of the newsgroups and mailing lists.
In short, how you wish to implement the system probably won't work.
Last summer I provided a standard bug reporting system (still living in
the file /pub/linux/scripts/ibug.tar.Z on tsx-11.mit.edu) for people
to send in bugs and have them processed automatically to the newsgroups
and mailing lists. Virtually nobody used the system: the unstructured,
distributed bug-reporting system that we have now actually works BETTER,
and here's why:
- Most people will report what they think is a "bug", when it's
actually an FAQ or a very common problem, or caused by their
own mistake.
- If bug reports are coallated and centralized, there will be
much less interaction between the developers and those who
find the "bug". As it is now, everything is posted to the
newsgroup or mailing list. This allows everyone reading to
see the problem, to know about it, and possibly offer solutions.
This information on "problems" with Linux isn't distributed
well enough, although it appears to be more organized.
- The current system lets anyone be involved, and help, and it's
no big deal if they goof. Any kind of official bug reporting
system will not only be too difficult for most people to use
(most folks can't even remember to include a Keywords line in
postings to c.o.l.a), but it will also intimidate those who have
bug reports and those who wish to help, but aren't sure if their
information is correct.
Linux has carried us quite far with the current "chaotic" widely distributed
bug-reporting "system". In fact, in recent months the number of so-called
"dumb questions" has gone down by about 500%. I would say that trying to
formalize a process which already works isn't a good idea.
Fred van Kempen and I have reached an agreement to work on a project
which will filter mails to the developers of various parts of the kernel
and other software for Linux. This will allow the developers to have
such mail automatically screened by program and by other volunteers so
they can spend more of their time writing code. It's not meant as a
replacement for the current newsgroup/mailing list paradigm, but instead
meant to augment it: a more direct route to discussing problems with
developers, while keeping the signal to noise ratio at an optimum level.
I'll be happy to provide more information or discuss this further.
I'd like to help with any such effort, however, it's important that we
decide just what that effort should be.
Cheers,
mdw
-- Matt Welsh, mdw@tc.cornell.edu GCS GMU d--- f- p--/-p+ c+++ m-- s- g++ "That kernel's got a mean streak A MILE WIDE!!" --Chip Salzenberg