Re: [Rosegarden-devel] MIDI-clock, port routing

[Pedro Lopez-C. <pl...@te...>]

On Tuesday 03 June 2003 13:26, Chris Cannam wrote:
> Richard Bown wrote:
> > - SUBS flags needed for the kaconnect entries to appear at all for an
> >   ALSA client.
> > - setting NO_EXPORT will hide them from kaconnect.
>
> Uh, did you intend some difference in meaning between "appearing at
> all" and not being "hidden"?
>
> iow, are you saying that omitting SUBS and adding NO_EXPORT will have
> the same effect: removing the ports from kaconnect (in the case of
> SUBS presumably only removing the port from one side at a time,
> NO_EXPORT removing everything)?  Or does one deal with the port
> names and the other with presence/absence of the connecting lines
> or something?

It would be better to not take my opinion very seriously, when i'm talking 
about the meaning of some ALSA flags. You remember why ;-)

Said that, this is what i think about this bussiness.

SND_SEQ_PORT_CAP_NO_EXPORT is not mandatory. Kaconnect needs to check 
explicitly this capability and if set, avoid to show the port. BTW, aconnect 
(the command line utility from alsa-utils)  ignores this flag. It shows RG 
ports and its connections. Of course, you can modify kaconnect to ignore this  
flag and she will gently show you the "hidden" connections.

SND_SEQ_PORT_CAP_WRITE / SUBS_WRITE and READ / SUBS_READ are very different. 
These flags are like the Unix permissions for files, but acting for programs  
instead of users. CAP_WRITE and CAP_READ flags alone means that only the 
program owning the port can establish a connection to/from the port. 
SUBS_READ and SUBS_WRITE means that any program can make subscriptions. For 
example, aconnect shows that RG (client 128 port 0) has a connection with 
64:0, but you will be unable to break this connection with 
	$ aconnect -d 128:0 64:0
and there is no way to hack aconnect to allow this.

Regards,
Pedro

-- 
ALSA Library Bindings for Pascal
http://alsapas.alturl.com