Comment 4 for bug 1521172

As mentioned by Adam, there is some contract pressure to ship a newer version of glibc than 2.21, which is now a year old. It is also better for the LTS in general if we can have this newer upstream version of glibc. However, as always, that needs to be balanced against the risks.

The risks are mitigated here by the fact that we have extensive autopkgtest coverage for the reverse-dependencies of libc (aka, "the archive"), and that proposed-migration gates on autopkgtest regressions for 5 of 7 release architectures. The remaining two architectures are powerpc, which is a best-effort community port; and arm64, which Canonical accepts responsibility for finding and fixing regressions on before release.

I understand that we will also have an archive rebuild test between now and release. It's unfortunate that this glibc update has not been included in the most recent rebuild test, but as long as we are still able to get one before 16.04 release that should suffice.

Along with some basic boot smoketesting prior to release from xenial-proposed, I believe this provides adequate coverage for us regarding risk of regressions.

So +1 from me for the FFe.