Re: Preparing (lib)gnome-keyring for the freeze.

[Steven Chamberlain <steven@pyro.eu.org>]

On 08:47, Andreas Henriksson wrote:
> To sum up the current problem, the versions in unstable are RC-buggy
> because of FTBFS. The FTBFS is caused by a newly introduced testsuite
> which basically showed the package never worked on kFreeBSD.

I assumed the package would still work;  unless I overlooked something,
the code contains a fallback to regular malloc in case mlock() is not
possible?

The failing test has code to check that it actually did use mlock().
Making the testsuite failure cause FTBFS of the package on kfreebsd
therefore seems... premature.

This did expose an important issue though - mlock() may be not working
yet on kFreeBSD, even on a sufficiently recent kernel where we'd expect
it to.

Time-permitting, we'd surely like to fix it for jessie, and perhaps we
will.  But I personally don't see it as a release-critical feature since
it isn't a regression and is an optional part of POSIX.

There may be some new issues during build/testsuite now as mentioned by
Bastian.  If so, I'd probably look into those first.

Regards,
-- 
Steven Chamberlain
steven@pyro.eu.org