16 Apr
2010
16 Apr
'10
3:30 p.m.
On Fri, Apr 16, 2010 at 10:15 AM, Adrian Knoth
<adi(a)drcomp.erfurt.thur.de> wrote:
First, we can't have virtual packages for shared
libraries in Debian, so
we cannot provide two different versions of libjack.
i don't understand this. either i'm not understanding the point, or it
sounds likea debian-specific limitation. i use fedora+ccrma, which has
jack2. i removed jack2 (--nodeps) and overlaid jack1. all fedora
packages that relate to JACK continue to work normally. am i not
understanding what you mean?
Second, we don't want application A require jack1
and application B
jackd2, so you can't run both applications. As this would be the case if
you have different jackd versions with different feature sets, we
entirely counter this by only packaging one version.
this seems entirely reasonable.
This is also beneficial wrt supporting. We don't
want different problems
with different versions, we don't want writing patches twice, once for
jackd1 and a second time for jackd2.
this is understandable, if less reasonable. at this point in time, the
bugs in jack2 are entirely different bugs from the bugs in jack1. the
existence of a bug in either version doesn't predict the status of
that functionality in the other.
Foremost, we don't want users to stumble because
they're using "the
wrong" version. To us, jackd has nothing to do with choice, it's simply
an inter-application framework for routing audio/midi data, and no user
should ever need to care about this. (Do you care about libreadline? It
simply has to be there)
this seems sort of right, although only as long as you use "jack" to
refer to the library+server and not any of the front end control
apps,patchers etc.
That said, we expect upstream to provide at least one
feature-complete
jackd implementation. This means DBUS support (pulseaudio integration),
jack-session support, ladish support or whatever the feature should be,
e.g. SMP.
this is where things really fall apart because it presupposes
agreement on the feature set, an agreement which as i think you know
just isn't there.
Since jack-session is rather new and the decision was
made prior to
this, since CCRMA and Gentoo's pro-audio overlay all use jackd2, we
chose jackd2 to be this feature-complete jackd implementation.
Consider this as a chance to avoid even more
divergence among jackd
implementations.
I understand the sentiment, but I am not sure that this is likely to
have this result.