Bug#704215: unblock: resiprocate - built against newer boost-dev
On 29/03/13 17:56, Axel Beckert wrote:
> Hi,
>
> Daniel Pocock wrote:
>> Please confirm if I should do the upload with a new version number
>> (1.8.5-4),
>
> You seem to have uploaded that already, but I haven't seen any
> confirmation on debian-release for that. Despite...
>
>> or if there is some other way to force it to be rebuilt for
>> the existing version?
>
> ... there is a very common way to do such rebuilds called "Binary NMU"
> or binnmu for short. Why didn't you wait for the confirmation you
> requested?
Axel, I'm sorry for the misunderstanding, I didn't realise that a
positive confirmation was always needed to upload to unstable. Due to
the holiday weekend, I won't be spending a lot of time at my desk and
did the upload proactively so that the issue can be fixed and not hold
up the release.
I believe the binNMU would also be suitable in this case, as there is no
change to the source or packaging artifacts. I'm quite happy for the
release team to prefer a binNMU for 1.8.5-3 and not unblock 1.8.5-4.
On 29/03/13 17:37, Evgeni Golov wrote:
> On 03/29/2013 03:22 PM, Daniel Pocock wrote:
>> I've found that one of the binaries, resiprocate-turn-server, needs to
>> be built again with the latest gcc and boost-dev or it fails to run
>
> Why was/is this not handled by a binNMU? Are there other packages
> (theoretically) affected?
As for the impact on other packages: I would say there is a definite
possibility that packages would have this bug if the following
conditions exist:
a) package compiled with gcc 4.7
b) libboost-dev < 1.46 was present on the build machine at the time of
compile
The problem would only be observed in a binary when it first tries to
use a thread.
- reTurnServer starts a thread at startup, so the problem is noticed
immediately. Binaries that only start threads some time later may not
stop so obviously.
- reTurnServer has a catch-block that catches exceptions and logs an
error. Binaries that don't have such a catch block may appear to just
stop mysteriously without any immediate clues about the problem.
Reply to: