Re: [Jack-Devel] jack_bufsize callbacks during jack_activate

On Sun, Feb 13, 2011 at 07:49:18PM -0500, Paul Davis wrote:
> On Sun, Feb 13, 2011 at 4:54 PM, Arnold Krille <[hidden]> wrote:
> > On Sunday 13 February 2011 22:19:02 torbenh wrote:
> >> On Sun, Feb 13, 2011 at 10:10:42PM +0100, Arnold Krille wrote:
> 
> > Special sysex midi-ports could also be a use-case. Or some future control-
> > protocol.
> 
> the JACK MIDI system was specifically designed to exclude large sysex
> messages from the "design goals". however, the recent addition of a
> command-line specification for MIDI buffer sizes somewhat takes care
> of this (albeit in a somewhat wasteful way if only 1 port is ever
> going to deliver huge sysexes).
> 
> port sizing is port type specific, so a new control protocol can
> provide its own buffer size, either as an absolute size or as a
> multiple of the audio buffer size.
> 
> > The question is: Do we/you want to think ahead that far and make the buffer-
> > size not port-type specific but port-specific?
> 
> I return again to the basic philosophy of JACK's design: we are not
> trying to make all things possible. The goal is to provide a system
> that makes it as easy as possible to accomplish,oh, lets say 90% of
> what anyone could possibly want to do with anything remotely similar
> to JACK, and leave the last 10% (mostly corner cases) for skilled
> programmers to implement themselves to avoid over-complicating an
> already rather complex system (internally) and also to avoid adding
> complexity to what is, at core, a very simple API.

this does not cover the initial question.


-- 
torben Hohn

1297649671.28770_0.ltw:2,a <20110214021412.GH16542 at siel dot b>