Subject: Re: [linux-audio-dev] App intercomunication issues, some views. GSOUND ????
From: Paul Davis (pbd_AT_op.net)
Date: Thu Aug 01 2002 - 23:13:24 EEST
>> ardour may well end up as the codebase for a bunch of libraries, but i
>> don't think you could create ardour from a bunch of libraries.
>
>Well I have created a Rebirth 338 clone, can you create it ?.
>To make a correct GSound design I must begin on more general and easy apps.
>(Stage 1,Simple Editor-Rebirth).
Thats where I disagree. The "more general" apps are the big behemoths
like Nuendo. They require extremely complex designs if they are to
work well. They often require totally different internal design than
the small apps.
There are a dozen or so small "rebirth-ish" clones floating around for
linux. There are no multitrack, multichannel audio sequencers (yet;
several are under development, and have been for several years). Why?
Because the code designs used for the small programs don't scale in
any way to the big ones. The way Ardour has been written has been
entirely intentional so far. Its driven by the fact that we have no
useful source code base to look at for this stuff. We're all
developing it in the dark so to speak, unlike kernel developers, web
server and browser authors, and language implementors, who can
frequently look at other code as examples of how to do something.
Consequently, I've aimed to include roughly working versions of as
much functionality as possible early on, to be sure that the basic
design can handle it *all*. Can the disk i/o system handle 32 tracks?
at 96kHz? can the transport mechanism handle auto-looping? can you
punch in while playing? can undo/redo be designed in a way that makes
it easy to undo/redo anything in an efficient and fast manner? what
happens when there are hundreds of "regions" displayed on the screen?
can you respond to MTC slaving? does your GUI handle control via MMC
properly? and so on and so forth.
in the future, it hopefully won't be necessary to do this. instead,
you can go and look at rosegarden, muse, ardour, ecasound etc. and ask
"how do they do it?" and then start from there. but right now, nobody
really knows how to do all of this stuff unless they work at
digidesign, emagic, steinberg or one of the few smaller companies that
write such software.
by writing "small" programs first, you end up with a codebase that has
to constantly be radically overhauled as your expectations grow. i
hope that when ardour reaches version 1.0, the codebase that relates
to audio will not change very much at all (version 2.0 will include
MIDI recording+sequencing support, and so there will be much new code
to handle that).
--p
This archive was generated by hypermail 2b28 : Thu Aug 01 2002 - 23:20:33 EEST