File bugs, don't fix 'em.

John Gilmore gnu at
Wed Jan 7 05:48:39 EST 2009

genesee wrote:
> > before the mesh search cycle is done, the 'looking for a mesh' keeps
> > blinking forever while my wifi is solid. If I let the mesh search run it's
> > course, (while I'm doing something else takes a few minutes), then
> > there's just my solid wifi circle in my frame.
> That sounds like a bug (the perpetual blinking).  Manual intervention
> must be preventing some signal from getting through; we should
> definitely fix that so the states of both devices are consistent.
> File a ticket? :)  Thanks.
> - Eben

Why would anyone bother to file a ticket about this?  One was filed
*15 months* ago, reported "fixed" when it wasn't actually, reopened
three months ago, and is still not fixed.  See .  You yourself didn't look for an
existing ticket before suggesting that this user, who was nice enough
to help you debug your product, file yet another one.

The same bug has likely been reported in #5459 (13 months ago), #6872
(6 months ago), #6944 (8 months ago), and almost certainly more often
(I got tired of wasting my time looking).  None of these bugs has been
altered in any way in the last *THREE MONTHS*.  It's common knowledge
among users that there's a bogus mesh icon that blinks sometimes.
Everyone has learned to ignore it, including those responsible for
fixing it.

It seems to take special chivvying to get almost any bug report looked
at by a developer.  (I've filed a lot of bugs and chivvied a lot of
developers.)  It seems to me that in programmer culture, when there's
nobody measuring progress by bugs actually fixed over the last week or
month, generally bugs don't get fixed.  This was as true at Cygnus as
at OLPC (the difference is that significant Cygnus revenue came from
top-notch bug support, so mgmt paid a lot more attention to it; we
wrote the first free software bug tracker, PRMS, for example).  Here's
*everything* that happened in TRAC since before the New Year:

The latest XO release 8.2.0 has major bugs that make the whole machine
lock up; that make it impossible or painful to use with some encrypted
WiFi AP's; etc.  The memory lockup is a *regression* that didn't exist
in the prior release.  Since fixes for these are not going into the
current priority release (8.2.1), nobody is bothering with them.  Only
when it's the last week of the release will anybody bother looking at
bug lists(*) -- and then it'll be obvious that the release can't possibly
go out so broken, so "the schedule must be slipped".  Well, no, the
time to fix the big bugs was over the last month and over the next
month.  At least we have half that time left.

Eben, going to fix, not just file, any bugs this week?   :-)


(*): Exceptions: cjb, dsd and dsaxena have pushed along some fixes for
big longstanding power-related bugs in the last month.  Tho we have
not improved either battery life or our ability to measure it -- at
all -- over 8.2.0 yet, with only a month to go.

More information about the Devel mailing list