>To make sure the GSound API has the needed features, we are working also on >diferent applications, who help us to desing, detect/fix bugs, and redesing >the API if needed. > >NxsTools: Something like SoX. >NxsAudio: A single audio editor (Similar to Soundforge) >NxsrRAck: A rebirth 338 clone >NxsrDJ: somthing like Tracktor DJ. > >And perhaps some day apps like Nuendo and Reason.
i don't believe that your approach will work. as the author of the only vaguely nuendo-like program for linux, i don't believe that a design based on stuff like: ---------------------------------------------------------------------- GsDev: low level access to any native/virtual audio/midi system. (--Portaudio--/--PorMidi-) GsData: a subset of tools to manage usual audio/midi data. (???) GsCda: low level access to CD-RW/DVD hardware. (--CdParanoia--/--CdRecord--) GsAmod: A "modular data flow graph filter" aproach. (???) Gsff: A set of tools to make I/O from diferent File Formats (Libsndfile, libaudiofile) GsDsp: Common DSP operations. (Csound/Mustajuuri/MPEG4) ---------------------------------------------------------------------- can be turned into an efficient, easy to use system. Much of what is necessary is in the GsData category, and if you take a look at Ardour, you'll see that things get really hairy, really fast. there is a lot of code in libardour to support extremely important functionality that just doesn't show up in many other programs. just today, i had to change a deep and fundamental aspect of how things work internally to the disk buffering mechanism to support ardour's scrubbing mechanism, because scrubbing (notably supporting instant-reversal when you're scrubbing 16 tracks at one time) radically changes the assumptions i had used about the relationships between disk and audio I/O. its not that i think that its impossible to write general purpose libraries for some or even all of this stuff. its more that i think you can't do it without first writing a really big and complex application to fully understand all the performance and design issues. 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. --p
