Subject: Re: [linux-audio-dev] LADSPA and run_adding()
From: Paul Barton-Davis (pbd_AT_Op.Net)
Date: Sat Nov 18 2000 - 22:22:45 EET
>This is just what I meant. For some cases (like above), run_adding()
>wouldn't be any faster (if not the opposite), while in other it would be
>optimal. So the questions are, a) which is more natural from interface
>POV, run_replacing() or run_adding(), and b) which scenarios are more
>common (weighted by their complexity -> no need to optimize simple
>setups).
I agree.
>I agree that run_adding() is optimal in certain
>situations, but I think run_replacing() is generally more usable. This is
>why I think mandatory run_replacing() + optional run_adding() is the best
>combination.
Given that one or the other is the best in some *common* situations
(though I may agree with you that run_replacing() is more common that
run_adding()), and given that there isn't very much work to be done in
the plug to supply both, why not require it ? the argument is that the
host can do it, but that is suboptimal because of the L1/L2 cache
situation. the plugins get slightly larger, but the difference between
run_replacing() and run_adding() is so small that almost anyone can
code it. in fact, steve's xml parser might even be able to generate the
code automatically. too bad you can't pass "pointer to machine
instruction" in any known language :)
which reminds me, why don't we rename run() as run_replacing() so that
we can always be clear on whats going on ?
--p
This archive was generated by hypermail 2b28 : Sat Nov 18 2000 - 22:58:44 EET