M-.-n wrote:
Even tho the approach is not 'out there' (as
it seems quite a bit of
linux programs use this client / server approach) I have the feeling
it's the kind of details that could pull off users from adopting linux
as a music platform. If you plan on trying a synth, you'd want to see
it's face right ? It's kind of the same as the jackd/jackstart/qjackctl
issue. When I tryed it first, I got really confused by all of them.
Actually, I still didn't get the jackstart/jackd difference, my eee
seems to need jackstart, my ubuntu install doesn't even have one.
In my views, a single executable would be a lot better, and if anybody
wants to run it without a GUI, give them a command-line option. Sure
that will mean your main executable will depend on user interface libs
but i don't see anybody wanting to run the synth without wanting to
tweak it at some point so the libs will be needed anyway....
Respectfully, I
disagree. The client/server architecture is one I wish
more synths would adopt. Thanks to this design, the engine may be
controlled by MIDI or OSC, can be run with or without a (possibly
custom) GUI, and allows the engine to be used for purposes other than
those originally intended. The client/server distinction confers great
flexibility, though it may not be immediately apparent to the new user.
Linuxsampler works this way, as does SuperCollider3, and I believe the
new Csound API permits similar operation.
Best,
dp