RE: Test cases that are possible candidates for a waiver
[[ I got a very strange bounce from linuxbase.org when sending
this the first time. My apologies if this ends up a duplicate
for anyone -- mats ]]
Thanks, Andrew. That's all true... Nonetheless, informally,
there are some failures that "matter more" than others. Some
indicate a failure of a facility under expected operating
conditions, others of a failure under "boundary conditions"
(like the pathname length issue recently discussed) that are
indeed spec violations but maybe not a big problem. I'm being
lazy here and asking if anyone else has looked in detail at
these failures - as I said, I'm most worried about async I/O
and about the mmap/mprotect/... family.
My experience is that waivers tend to be kind of permanent.
{Do we have} / {should we define} a category for temporary
waivers - a recognition that current GNU/Linux code generically
has these problems, and the waiver is valid for 12 months only?
(12 months is an example, of course)
Mats
-----Original Message-----
From: Andrew Josey [mailto:ajosey@rdg.opengroup.org]
Sent: Wednesday, December 05, 2001 2:11 AM
To: Wichmann, Mats D; Kevin Caunt; ajosey@rdg.opengroup.org;
lsb-test@lists.linuxbase.org
Cc: lsb-test@linuxbase.org
Subject: Re: Test cases that are possible candidates for a waiver
The majority of these tests are known to pass on other commercial
operating systems (such as Solaris)--there are probably some test bugs
in there that would only show up on Linux -- but it does indicate some
mismatch between the current semantics and the standard
semantics for these interfaces. The test assertions are based on
standard specifications of the interfaces (Single UNIX Spec V2 and
gABI in some instances).
regards
Andrew
On Dec 4, 5:11pm in "RE: Test cases that ", Wichmann, Mats D wrote:
> This is similar to the list of suspects I've seen in limited
> testing. The question I have is, if these are waivered, do
> we still have a useful async I/O and memory mapped I/O
> implementation (those are the two areas that worry me, at
> first glance)?
>
> I'm not in a position to provide an answer to that question -
> so it's looking for more informed input. Are those testing
> boundary conditions or core functionality?
>
> Mats
>
>
>
>
>
> --
> To UNSUBSCRIBE, email to lsb-test-request@lists.linuxbase.org
> with subject of "unsubscribe". Trouble? Email
listmaster@lists.linuxbase.org
>-- End of excerpt from Wichmann, Mats D
--
To UNSUBSCRIBE, email to lsb-test-request@lists.linuxbase.org
with subject of "unsubscribe". Trouble? Email listmaster@lists.linuxbase.org
Reply to: