[LAD] Feature requests: add JackSession support

Paul Davis paul at linuxaudiosystems.com
Sun Jul 3 12:10:09 UTC 2011


On Sun, Jul 3, 2011 at 7:43 AM, Fons Adriaensen <fons at linuxaudio.org> wrote:
> JSM is one of them, it's a mess
> even in its initial form.

its important to point out that this is a criticism that could be
levelled at JACK too.

your analysis was thoughtful and sensible. as it turned out, neither
torben nor I agreed with it. as it stands, the JACK Session API is a
minimal, barely intrusive addition that gets done 90%+ of what the
users of JACK wanted/needed. the design goals behind JACK have always
included a strong aversion to going after the 100% solution because I
and others have strong feelings about the infeasibility of it in
general, mostly because of a profound awareness of the total SNAFUs
that seem to happen everytime somebody tries this in software. so no,
it doesn't do everything possible, or everything desirable, or
everything perfectly, but it does do most of what people actually
need, without having much impact on developers' own internal
architecture.

the lack of quit-without-save is a decision that i was opposed to, but
accepted precisely because it has the potential to interfere with
developer's internal architecture. most applications these days ask if
you attempt to quit without saving (some will even treat the window
manager close button as a trigger for this). quit-without-save
(potentially) requires a new code path that circumvents whatever
internal logic the program has to decide whether to ask the user. its
not a huge change, but it is a change that isn't merely additive (as
is the case for the rest of the API).  in the interest of seeing
something adopted that performs a useful role for users, it seemed
better not to require supporting JACK Session to also require "your
program must be able to quit without any user interaction".



More information about the Linux-audio-dev mailing list