these are all good observations and they are the types of things that need to be
taken care of in our 2.0 audio pipeline rewrite.



> So, whilst working on a mod playing lmc, I've come across a couple of things
> that are hindering my progress..  Everything should be relatively easy to fix
> or make work, but I figured I should ask for input before doing so, in case
> there were reasons for things being the way they are..
> Numero uno: mp3ism's outside of the xingmp3 lmc.  This includes things like the
> PMIs trying to read a id3 tag, the PMOs basing their time info off the number of
> mpeg frames the lmc sent, etc..  These (and probably other things I haven't
> noticed yet) should all be moved into the lmc so that the only thing that is
> file format dependent is the lmc.
> #2: This may be just my poor reading of the code, but... The current PMI/LMC/PMO
> interaction seems geared toward the grabbing a chunk of data, letting the
> lmc perform magic on it, and giving it to the pmo.  This works great for
> streamable formats, but really isn't ideal for non-streamable things like mods..
> With a mod, you really need the whole file available before you start generating
> the audio from it.  Hmm.. I suppose it'd be easy enough in the case of a mod
> player just to have the lmc tell the pmi to read all the data from the file
> before starting to do anything, but I see that as being a small waste of
> memory...  I'm not completely sure, but having the option to ask the pmi to do
> no buffering and just pass back a file pointer might be a good thing in this
> case..
> And on to something completely unrelated.. continuous playing of files.
> Currently, the pmo is deleted after every file in a playlist.  This is bad. =)
> It makes an audible click on my system, and means you can't play files without
> a small break between them..  Easiest way around this is to make the pmo
> persistant across songs/lmc/whatever.  It really only should fully quit when the
> audio format being played has changed, or the user presses stop.
> Comments? Code is forthcoming. =)
> Isaac

Reply via email to