Re: [Jack-Devel] jackd1/jackd2: FFADO port name divergence

Paul Davis <[hidden]> writes:

> On Wed, Jan 25, 2012 at 6:23 PM, Nedko Arnaudov <[hidden]> wrote:
>
>> Time will tell if the metadata api is the proper way. ATM in ladish I'm
>> using well known guids that I encode in room port names, but this is
>> only one of the other possible ways.
>
> useful/pretty/semantic names need to be accessible to any JACK client
> that wants to find stuff or display stuff. thus they need to have
> well-defined URIs that are not the domain of a single application.

Yes, we agree here I think. My point was about the implementation. uuids
are URIs too. IMO adding some lv2-style stuff to jack is a good thing.
lv2-styles stuff like use of real char string uris and extensions.
RDF/Trutle in jack is not welcome though.

> when a user makes a port into "monitors/L", they presumably want all
> apps to be able to show that as "monitors/L" even if the canonical
> name continues to exist.

Who will set tags like "monitors/L"? And how they will be preserved
across jack lifecycles? qjackctl or some other full session manager
like ladish can do this. Do you foresee something else instead?

Before adding more complexity to jack it would be useful to know the
practical benefits. From the ladish POV it makes sense to have small and
flexible jack core. <rant>There are some properties that are actually
making jack quite the opposite. Like client autorenames, lack of [good]
separation between process and client objects in the API and sacrificing
human readable names in favour for well-known "identifiers" that
actually benefit only users without multichannel hardware.</rant>

-- 
Nedko Arnaudov <GnuPG KeyID: 5D1B58ED>

1327535915.22670_0.ltw:2,a <87boprpc7q.fsf at arnaudov dot name>