Re: [Jack-Devel] lib{j, tsch}ack{, 1, 2} [Was: Jack "capsule/container" implementation?]

On Thu, Feb 10, 2011 at 09:35:51AM -0600, Gabriel M. Beddingfield wrote:
> 
> 
> On Thu, 10 Feb 2011, Paul Davis wrote:
> 
> >>...except that, as soon as they switch to Jack2 Debian had a JACK ABI
> >>breakage with a few apps.
> >
> >is this reported anywhere? did they bring it up with us? has anybody
> >got any clue what was going on there? was it actually ABI breakage or
> >bugs in one or the other implementation of JACK? these are entirely
> >different things and should be handled differently (by us, and by
> >packagers).
> 
> Yes, I reported it here as soon as I saw the bug report in Debian.
> Probably more of a "bug" because Jack2 was tried to supply a symbol
> (don't remember which, an obscure function call), but for some
> reason it wasn't making it into the library.  I didn't follow
> through the resolution.
> 
> Still, after spending a lot of time with Debian, making strong
> assertions about the ABI... I felt pretty stupid when these bug
> reports came up.

this is like claiming some software is bugfree, and then finding a bug.
we always said, that if ABI incompatibility shows up, its a bug.

and we normally react pretty quickly, when its raised on the ML.
the real problem are the different thread semantics of 
jack1 vs. (jack2 tschack)

these are very subtle, and poorly documented.

-- 
torben Hohn

1297424618.29553_0.ltw:2,a <20110211114304.GK5550 at siel dot b>