[pkg-wpa-devel] Bug#806889: Bug#806889: wpasupplicant: New upstream version (2.5)

[Faidon Liambotis]

On Sun, May 01, 2016 at 07:04:56PM +0200, Stefan Lippers-Hollmann wrote:
> > Moreover, there are several open bug reports that may have been fixed by
> > a new version (e.g. #815121) or can be easily fixed or closed (e.g.
> > #729934).
> 
> #729934 will indeed by fixed by the next upload/ binNMU, thanks to the
> new dbgsym packages introduced by dh.

Indeed — but the point was that the bug queue for wpa seems to be
largely untriaged. #729934 was just an example of a 2½-year old bug
report with no response whatsoever, that's even now at a stage that
basically now needs zero effort from your side to fix, due to toolchain
changes.

> I'm still debugging the 4addr regression mentioned in my previous mail,
> given the rather unfortunate approach to reproduce it, testing takes
> about a week for each iteration (unless it fails early by chance, so
> far I've seen time to failure up to 3-4 days, even though it often 
> within hours). Replacing one bug with another doesn't sound like a good
> tradeoff to me.

Sure it does.

This regression has been reported by only you, occurs only in very
specific circumstances (both wpasupplicant and hostapd, hard to
reproduce as you yourself say), and hasn't been a blocker to the rest of
the Linux world carrying this new upstream version since it was
released, more than half a year to a year (depending on whether it's a
2.5 or 2.4 regression). It hasn't been such a big deal to upstream
either, since they haven't released an updated version fixing such a
bug. Clearly not a very important bug then, no?. (plus, are you really
working on it? it's been a month and a half since you last said anything
on this bug report; have you contacted your upstream?)

On the other hand, we have *two* major new upstream releases that will
close at least a few Debian bug reports, plus incorporate all kinds of
fixes and new features that upstream has implemented in the year between
2.3 and 2.5. Such an example would be Network Manager 1.2's MAC address
randomization while scanning, a pretty important feature from a privacy
perspective.

Thanks,
Faidon