Re: lsb-fhs test suite failures
hi Chris
Since this would affect certification I think a LSB Certification
problem report needs to be filed . My reading of the
FHS is that is a requirement, since it makes no sense to
to have a negative recommended practise (should not), but does
make sense to read that as "shall not". Whether or not that is sensible
is another matter.
By having a problem report on the record, we can get an official resolution,
and record the issue if it is a specification issue (which I suspect
it is). I'd recommend filing it as a spec problem in the gLSB (which
refers to the FHS), you can use
the lsb-runtime for IA32 as the test suite, but note that the problem
is being encountered on other architectures but could apply to IA32 systems
potentially also.
Btw is the allocated devices doc intended to be architecture neutral?
regards
Andrew
On Jul 18, 5:23pm in "lsb-fhs test suite f", Christopher Yeoh wrote:
> Hi Andrew,
>
> On some systems (eg Debian PPC32) the following tests fail:
>
> /tset/LSB.fhs/linux/dev/dev-tc 23 FAIL
> /tset/LSB.fhs/linux/dev/dev-tc 24 FAIL
>
> These two tests check that the major and minor numbers of /dev/vcs1
> .. /dev/cvs12 and /dev/vcsa1 .. /dev/vcsa12 are correct. On
> the system being tested these are actually symlinks. Eg.
>
> lr-xr-xr-x 1 root root 5 Jun 14 16:52 /dev/vcs1 -> vcc/1
>
> The test currently fails because it checks the symlink and not what it
> points to.
>
> In general in the FHS symlinks are ok as substitutes, but in Section
> 6.1.3 of the FHS it says:
>
> "Symbolic links in /dev should not be distributed with Linux systems
> except as provided in the Linux Allocated Devices document."
>
> though taking the definition of "should" from the LSB specification:
>
> "For an implementation that conforms to this document, describes a
> feature or behavior that is recommended but not mandatory."
>
> As an aside the future direction of the FHS appears to be to remove
> this statement from the normative part of the document.
>
> So it seems to me that it would be ok to change the testing of device
> files so that if a symlink is encountered the tests check the target
> of the symlink rather than failing. Any comments?
>
> Regards,
>
> Chris
> --
> cyeoh@au.ibm.com
> IBM OzLabs Linux Development Group
> Canberra, Australia
>-- End of excerpt from Christopher Yeoh
--
To UNSUBSCRIBE, email to lsb-test-request@lists.linuxbase.org
with subject of "unsubscribe". Trouble? Email listmaster@lists.linuxbase.org
Reply to: