re: RasterMan on linux and threads

[Dan Kegel <dank@alumni.caltech.edu>]

re http://kernelnotes.org/lnxlists/linux-kernel/lk_9912_03/msg00480.html

Rasterman is wrong in saying that all threads run on the same
processor, but everything else he says about threads is spot on:
unless you have a real good reason to create a thread, don't.
Ousterhout certainly agrees.  See his slides on "Why threads are bad (usually)"
at http://www.scriptics.com/people/john.ousterhout/

Earlier in his news page, Raster says:
> The [imlib2] API changed a lot recently - it's now context based (you set
> the current context) and it uses that current context for almost all
> calls. It means many fewer parameters to many calls - but it also means
> Imlib2 is NOT thread-safe as the context is global. If anyone wants to
> work on making it thread safe - there's source code and send patches -
> but I don't see much value in it. Do all your graphics work in 1 thread.    

Wow, deja vu.  Right on the heels of Jim Gettys' post on the same subject,
http://kernelnotes.org/lnxlists/linux-kernel/lk_9912_03/msg00096.html

Raster, you might want to read Jim's post.  X calls use lots of context 
parameters (like old imlib), OpenGL calls use none (like new imlib2), and 
now they both wish they'd used a single one (i.e. an object oriented approach).

And you might want to modify your statement that threads always run on the same
CPU...
- Dan

-- 
(The above is just my personal opinion; I don't speak for my employer,
 except on the occasional talk show.)

-
To unsubscribe from this list: send the line "unsubscribe linux-kernel" in
the body of a message to majordomo@vger.rutgers.edu
Please read the FAQ at http://www.tux.org/lkml/