Comment 86 for bug 893091

Hi Jason,

On Jan 19, 2012, at 11:54 PM, Jason Gerard DeRose wrote:

>I'm trying to back-port this to Oneiric to deliver in the Novacut Stable
>PPA. This will allow us to maintain compatibility with Oneiric for a
>bit longer, and I figure help this shiny new port get more testing.

I think that's great. I see no reason why the port shouldn't work on Oneiric.
In fact, I started the work there, before I upgraded my machines to Precise.

>So my question is: do you think this is a reasonable thing to do, or is
>this too risky/invasive? Would having this tested on Oneiric be of any
>use in the first place?

I think it would be great to have in a PPA. I can't think of any risks. It's
not an SRU candidate though.

>If you think this is a bad idea, we'll drop support for Oneiric this
>month... porting the Dmedia DBus service to python3-dbus went
>wonderfully smooth, and I'm not turning back! Thanks for the fantastic
>port!

Awesome!

>BTW, I get some test failures when building on Oneiric, although I'm not
>sure if this is reason for much concern:
>
>https://launchpad.net/~novacut/+archive/daily/+build/3106062
>
>https://launchpad.net/~novacut/+archive/daily/+build/3106063

Yep, these are known, and actually caused by some buggy packaging of dbus. In
order to move the dbus binaries to /bin from /usr/bin, an unfortunate bug
crept into the packaging. The debug version of dbus-launch ends up in the
production deb.

The effect of that is that when dbus gets a warning and DBUS_FATAL_WARNINGS is
enabled, dbus-launch will abort. That's the signal 6 you're seeing. In
practice this doesn't seem to affect anything because that envar is never
enabled afaict. It *is* enabled in the dbus-python test suite, which is why
the builds are failing.

https://bugs.launchpad.net/ubuntu/+source/dbus/+bug/913991

It's fixed in Precise of course. :)

My suggestion is to comment out this line in dbus-python's
dbus/test/run-test.sh script:

export DBUS_FATAL_WARNINGS=1

Your PPA should then build just fine.

Cheers!