JACK and Pipewire experiences and latest issues

Hi all,

In the last months I had quite seamlessly switched to pipewire on=20
arch-based distributions (Manjaro and now CachyOS).

It seemed to 'coexist' quite well with jack applications via=20
pipewire-jack package "Low-latency audio/video router and processor -=20
JACK replacement".

Essentially from a user point of view this was quite similar to my=20
pulseaudio set-up with having jack "always on" and the pulseaudio-jack=20
sink which works for a one machine set-up.

Lately I've seen some issues and idiosyncrasies and I'm not sure what=20
these are related to if pipewire, applications updating, or both:

-=C2=A0 Ardour can't seem to export any more if 'jack' mode is on: i.e. i=
t=20
has to be set to Alsa in order to successfully export sessions. In turn=20
this means when Ardour is in alsa mode it exclusively takes audio and=20
cannot send/accept audio from other application (one of the advantages=20
and features of jack and - in theory pipewire too).

- Some applications like Yoshimi show qirky behaviour now in jack mode.=20
For example if the 'quantum' is set to lower e.g. 128 Yoshimi will start=20
and make not-nice noises. The work-around is to set Yoshimi to use alsa.=20
In this case it is shown in pipewire 'graphs'

- qjackctl seemed to happily work and show pipewire (audio) applications=20
as well... But not I'm not sure it's still relevant. qpwgraph seems to=20
be the successor but it seems it only has the 'patchbay' view which is=20
usually not my favourite, especially in certain situations

- Setting 'quantum' kinda works but I have instances where it seems to=20
get (re)set by applications. Plus it's still kinda officially only=20
command-line (cable supports it but I don't think it's in any official=20
repository yet, I made my own Python Tk 'gui' before knowing about=20
cable), maybe qpwgraph should offer the functionality as the 'successor'=20
of qjackctl?

I know this is very random stuff and probably most of the above should=20
be reported to either each application or pipewire. But I'm just trying=20
to gauge what other users' experiences are. Also it seems all of these=20
strange behaviours appeared in the last months so I'm wondering why=20
everything seemed to work pretty well before and what changes might have=20
occurred at a more systematic level.

Also I've already been through a rather lengthy and complicated bug=20
reporting loop where one application seemed to misbehave with pipewire=20
(midi) and seemed to point to a pipwire bug, then pipewire seemed to=20
have fixed the bug, but then it seems to be still there. I'm wondering,=20
given that the Linux Audio Community is probably not _that_ big how=20
efforts could be better harmonized :-)

I'm also sure I'm missing (as a user) a few nuances about pipewire and=20
it's design, and I think that's because lately I'm trying to focus on..=20
making music :-)

Lorenzo