Re: strace broken in 7.0?

From: Robert Watson <rwatson_at_FreeBSD.org>
Date: Thu, 10 Jan 2008 21:16:54 +0000 (GMT)
On Thu, 10 Jan 2008, Timo Schoeler wrote:

> So a 'conservative development model' (the BSDs where typically known for) 
> leads to a well functioning environment. The only BSD I know of where this 
> really *is* the case is OpenBSD. NetBSD, e.g., had times where their tree 
> didn't build (especially for me between 3.99.3 and 3.99.15 on macppc). 
> FreeBSD has a damn high number of open/unfixed PRs, and people complain more 
> and more that this harms them in several ways.

I think you make a number of important points, but I want to make one 
observation regarding PR count: the number of problem reports doesn't 
correspond directly with the number of bugs in the system, it also tracks the 
number of users of the system.  I don't think anyone is really in a position 
to directly evaluate the relative "bugginess" of various systems in an 
objective way, but I think I would probably be right if I claimed that FreeBSD 
has a lot greater installation diversity and end-user exposure than OpenBSD or 
NetBSD, and this is certainly a contributing factor to our bug report 
situation.  Derived systems like PC-BSD have only increased that exposure, 
especially in the desktop hardware space.

I know that Mark Linimon has done quite a bit of analysis of the state of the 
PRs, especially as to which ones stay open vs. which ones get closed, and may 
be able to offer some insight.  I have a vague recollection that last time 
around, he reported essentially linear growth in open kernel bug reports, and 
essentially stable ports PRs, but I've not really seen stats on how bug 
reports against the base system get closed.  For example, I'm not sure we make 
a "fixed" vs "closed" distinction, which we'd need in order to do a good 
analysis.

FWIW, there are quite a few of us who are interested in improving the 
daignosis of problems.  My recent work involving DDB textdumps was entirely 
about improving our level of automation for kernel crashes, which should allow 
us to do a lot less hand-holding in gathering bug reports (and, ideally, 
therefore improve our report handling rate).  While there are social elements 
at work here that aren't addressed by purely technical solutions, there is 
also a lot of room for the development of technical solutions here as well.

Robert N M Watson
Computer Laboratory
University of Cambridge
Received on Thu Jan 10 2008 - 20:16:58 UTC

This archive was generated by hypermail 2.4.0 : Wed May 19 2021 - 11:39:25 UTC