Addressing the pressing performance / bug issues (was "FreeBSD's
problems as seen by the BSDForen.de community")
Robert Watson
rwatson at FreeBSD.org
Thu Jan 10 13:02:24 PST 2008
On Thu, 10 Jan 2008, Garrett Cooper wrote:
> I've been reading the thread a bit (glossed over the licensing political
> drumming), and I realize that yes there is in fact a need for addressing
> issues. Would it be a good idea to compile a series of bugs using a
> prioritization system based on interest (number of stakeholders) and/or
> funding (say Cisco or Intel devotes X number of dollars for a feature)? I
> would think that that would be an effective system for gaging priority
> perhaps.
I suspect that of all the stakeholders in the community, the Ciscos, Junipers,
etc, of the world are the ones who will get their problems addressed most
quickly, on the basis that while OS developers are expensive, they're not all
that expensive. The people who tend to do least well are the individual
end-users who don't have the technical know-how to engage with developers who
have, on occasion, been known to have poor personal communications skills, not
to mention a day job. The most common classes of problems that tend to go
unsolved are the ones involving pseudo-random combinations of hardware not in
the hands of developers, less common configurations using networking
tehnologies that aren't as widely deployed, and aging hardware -- the precise
types of problems that big companies try to avoid (especially minimizing
hardware footprint).
It seems to me that we have structural, procedural, and social areas that need
addressing with respect to how bug reports are handled. We need to manage and
present them better (no one needs to rant about GNATS, we all know it leaves
much to be desired), and we need to build up a better social model regarding
how we handle reports. For example, how about we have a web page that shows
the most recent 50 GNATS bug reports in a single click? Or a way for the RE
team to flag potentially interesting issues for releases so that developers
(and end-users) can look for things to work on? The problems are similar to
problems regarding code review: how can we get people interested and engaged
in debugging problems, and how can we improve the tools to make it happen more
smoothly? I like the idea of Bugathons, we should see if we can't get more
people involved in that as well.
Finally, I think there's a set of people in the FreeBSD project who really
deserve our thanks -- that's the people (and there are several) who watch new
bug reports come into GNATS and attempt to sort them, triage them in various
ways. Turning "my machine crashed" into a usable bug report requires a lot of
dedication from these folks, and we should be doing whatever we can to support
and help them, including trying to find more volunteers to work on that
project.
Robert N M Watson
Computer Laboratory
University of Cambridge
More information about the freebsd-current
mailing list