5.16 packaging underway

[Dominic Hargreaves]

I made a start at packaging 5.16 for experimental at the weekend.
The results so far are in the debian-5.16 branch in the usual repository

http://anonscm.debian.org/gitweb/?p=perl/perl.git;a=shortlog;h=refs/heads/debian-5.16

The test build status is as follows:

i386 on squeeze chroot with ext3: fails due to RT #112866 (will be fixed
for RC1).

amd64: builds okay

kfreebsd-amd64: the following fail:

Failed 2 tests out of 2276, 99.91% okay.
        ../dist/threads-shared/t/waithires.t
        op/getpid.t

The former is presumably the same issue as #672152 which is due to
a recent change in eglibc.

The latter is related to a change made in February as a result of
https://rt.perl.org/rt3//Ticket/Display.html?id=96270 filed by Niko.
Even though

LinuxThreads is reported to be in use, the LinuxThreads specific test
fails:

op/getpid.t .. 1/3 # We're running under gnukfreebsd with linuxthreads <linuxthreads-0.10>
# Failed test 2 - getpid() in a thread is different from the parent on this non-POSIX system at op/getpid.t line 40
# it should not be "67271"
# but it is.
# Failed test 3 - getppid() in a thread is different from the parent on this non-POSIX system at op/getpid.t line 41
# it should not be "67270"
# but it is.
op/getpid.t .. Failed 2/3 subtests 

However: this only happens in the Debian build, not a vanilla build
of 5.16.0-RC0 configured with

 ./Configure -des
 make test

The difference presumably relates to

 -Dusethreads

for the Debian build.

With regard to the last point: any help welcomed, but solving #672152
should take priority.

I hope to be able to kick off some archive rebuilds to help provide
test coverage for upstream prior to 5.16.0, as well as possibly an
upload to experimental to get some architecture test coverage, but both
depend on having enough time during the week (and also at least RC1
being released).

-- 
Dominic Hargreaves | http://www.larted.org.uk/~dom/
PGP key 5178E2A5 from the.earth.li (keyserver,web,email)