[lsb-discuss] Resending tests for libc APIs

Andrew Josey ajosey at rdg.opengroup.org
Mon Oct 29 09:18:03 PDT 2007


hi Stew, All
Not sure these assertions are based on POSIX, in which case can you
ensure the test cases are only run when LSB_TEST=true or yes (i cannot
remember off hand) . This way the license terms can be observed, i.e.
the original test modes have to be preserved, so as not to cause
confusion if someone tests for POSIX compliance.
thanks
Andrew

On Mon, 2007-10-29 at 12:07 -0400, Stew Benedict wrote:
> 
> /tset/POSIX.os/files/posix_fadvise/T.posix_fadvise 3 FAIL
> 520|131 3 00016383 1 1|posix_fadvise returned Success instead of
> EINVAL
> 220|131 3 1 09:22:34|FAIL
> (passed -1 as advice, when manpage says 0-7 are the only valid values)
> 
> /tset/POSIX.os/files/posix_madvise/T.posix_madvise 4 FAIL
> 520|132 4 00016384 1 1|posix_madvise returned Success instead of
> EINVAL
> 220|132 4 1 09:22:35|FAIL
> 410|132 4 1 09:22:35|IC End
> (passed -1 as advice, when manpage says 0-4 are the only valid values)
> 
> It would seem the test method is valid, but the behavior, at least on
> this 
> platform does not agree with the man page. Did the tests pass on your
> test 
> platform?
> 
> Anyone else have any comments? 




More information about the lsb-discuss mailing list