Re: The case for FreeBSD

In message <4207F43F.3040300_at_fightevil.net>, njc writes:
>Scott Long wrote:
>> We need more people 
>> who will write articles and papers and do benchmarks and regression
>> testing.  That's not to say that we don't already have people filling
>> these roles, it's to say that we need more.
>
>Scott - 
>
>  Do you, or possibly other developers, have any suggestions on the best way to organize
>a community-driven testbed and quality assurance effort? Specifically - what would be the most
>convenient form and method for a BSD developer to receive feedback for patch testing, are there
>any recommended testing procedures (methods && tools), etc?

Here's what I'm looking for in a good bug report (in no particular order):

    Stack backtrace if the kernel croaks.

    Ktrace if the kernel does something wrong.

    Elimination of unrelated factors.
	If you see the error compiling ports/x11/xorg, try to find out how little of it
	you can get away with doing to provoke the error ?

	If you are using complex disk geometries, try if you can reproduce on a plain
	single disk system ?

	Does it happen on other computers as well ?

	Does it happen only under SMP or also UP ?

	Does it happen also in single user mode ?

	Does enabling WITNESS/DIAGNOSTICS in the kernel catch something ?

	Can you write a shell script which fails every time ?

	Can you find another way to provoke it ?

	A patch.

It is really very very simple, the easier you make it for me to reproduce the
error here, the faster I can find out what's wrong and fix it.

-- 
Poul-Henning Kamp       | UNIX since Zilog Zeus 3.20
phk_at_FreeBSD.ORG         | TCP/IP since RFC 956
FreeBSD committer       | BSD since 4.3-tahoe    
Never attribute to malice what can adequately be explained by incompetence.