Subject: Re: Regressing testing of system calls
To: Darren Reed <>
From: Giles Lean <>
List: tech-kern
Date: 06/15/1999 07:25:56
On 14 Jun 1999 12:51:53 -0700  Chris G. Demetriou wrote:

> If you want to do this kind of testing, personally, i'd think you're
> best off either starting with a POSIX (or other) conformance test
> suite, or at least a standards document, and going from there, and, if
> necessary, fixing up the manual pages as you go to match.

A first pass through the manaual pages and the Single Unix
Specification does show differences.  Mostly we document additional
returns, but sometimes the standard has extras.  (They're fond of
EINTR on just about anything!)

It might take a policy decision to judge whether the
standard-but-impossible returns should be added to our manual pages
but any validly found to be missing could readily be added. :-)

Given how hard it is to provoke error handling code sometimes I expect
code review will be necessary in determining what errors we can in
fact return.

Standards.  Hmm.