[gmpi] Re: Reqs 5, 6, 11 for debate

  • From: Paul Davis <paul@xxxxxxxxxxxxxxxxxxxxx>
  • To: gmpi@xxxxxxxxxxxxx
  • Date: Thu, 20 Nov 2003 16:10:05 -0500

>>>>
>so, you want us to pollute GMPI with extra junk just to to fit in the
>misdesigns of Redmond?
><<<
>
>No, definitely not.  But I do want to make sure GMPI plugins will be able to
>work in these RT contexts.
>
>>>>
>you shouldn't be doing any memory allocation or any disk io or any blocking
>on anything in the realtime audio thread
><<<
>
>Agreed.  But the memory touched by your realtime DSP code needs to have been
>allocated from this special pool.  And if your plugin does disk streaming,
>the disk IO calls need to come from a special place too.

I don't want to rant, but I just can't help myself.

What do these idiots at MS think they are doing? Do they think this is
some new problem that has never been seen before? That they are the
first to tackle it? That the guys in the R&D group (quite a few of
whom I knew back when they were graduate students) have come up with
something mysterious and new? No! Special pool? Oh, you mean that its
been pinned in RAM so that it can't be faulted? POSIX 1.b defines both
mlock()/mlockall() that do this with or without realtime or
reservation scheduling. RT scheduling? POSIX 1.b defines both
SCHED_FIFO/SCHED_RR policies, with a mechanism for adding other
policies too (hence the patches for linux to add a reservation
scheduling policy). there's no need for any new API for any of this
stuff - its been defined for ages. 

We all know, from years of experience, that the rules in a realtime
thread are very simple: no memory allocation, no disk or network i/o,
no blocking semantics of any kind. We've all become familiar with the
correct set of programming idioms (lock free data structures,
conditional locks, etc) that are needed to pull all this off.

I'm sorry, I am just really tired of goons who have nothing but
marketing and money at the center of their universe messing around
with designing and writing operating systems. There is no reason for
the approach that Longhorn is going to force on you Ron, you know it,
I know it and anybody with any OS-level experience knows it. Its
bullshit, and I'm tired of having to workaround MS' profit model for
technical designs. Actually, I'm not tired of it - I normally avoid it
completely.

OK, I'll stop now. Really. Honest. I'll try my best.

--p



----------------------------------------------------------------------
Generalized Music Plugin Interface (GMPI) public discussion list
Participation in this list is contingent upon your abiding by the
following rules:  Please stay on topic.  You are responsible for your own
words.  Please respect your fellow subscribers.  Please do not
redistribute anyone else's words without their permission.

Archive: //www.freelists.org/archives/gmpi
Email gmpi-request@xxxxxxxxxxxxx w/ subject "unsubscribe" to unsubscribe

Other related posts:

  1. Home
  2. gmpi
  3. Archive
  4. 11-2003