guile-devel
[Top][All Lists]
Advanced

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: More Bug Stuff


From: Thien-Thi Nguyen
Subject: Re: More Bug Stuff
Date: Mon, 25 Mar 2002 00:21:30 -0800

   From: Marius Vollmer <address@hidden>
   Date: 25 Mar 2002 00:03:00 +0100

   Are we talking about tasks here as well, in addition to bugs?  For
   tasks, priority makes more sense.

bugs can result in tasks (like fixing them), and not all tasks are
bug-related.  over time tasks that encounter bugs become bug-related.
presumably people might think it's a good idea to scan the bug list to
see which bugs relate to work in progress, and adding info to the bug db
whenever something interesting occurs might help fix bugs faster, and
DTRT.

so to me these two areas can be better managed separately.  bugs db
should be descriptive and avoid being prescriptive.  referring to a task
from a bug header is already supported.

task prioritization is undocumented mostly because everyone (human) is
motivated differently when they look at any set of tasks.  it is a
personal matter which some may never be comfortable in disclosing.  more
basically, even knowing one's own motivations is difficult sometimes.

probably we can come to rough consensus (as we go along, and enough to
get stuff done) at which time someone should write tasks/TRIAGE.  (a
bugs/TRIAGE is a good idea too.)  perhaps we can borrow further from
debian.

after some thought i believe one dir per bug is a better overall fit.
sorry about being overhasty.  every other approach invents weird syntax
(like the one we're leaving) to forget and screw up.  [jobs clones
guffaw "bundles" into their grey cubes.]

on the other hand, single file at top is simple and here.  so really, my
proposal (now -- please ignore previous) comes down to:

        N  -- the rfc822 for bug N
        .N -- the directory w/ all of bug N's related files;
              may be empty or absent

in the headers, this would add required header:

        bug-stuff-dir: .N

under .N things can be named in regular ways, such as "test-case-1.scm"
or "why-i-must-rant-against-this-bug" or "rfc822" (symlink).  the naming
of these files can be conventionalized later and w/o overmuch regard to
the actual bug number.  programs that munge these files are accordingly
lightened.  programs that munge bug sets need to stay informed w/
`bug-stuff-dir' but that's just another header (already done).

thi



reply via email to

[Prev in Thread] Current Thread [Next in Thread]