Re: A synchronization module, + thread-specific 'import' bug

Guido.van.Rossum@cwi.nl
Tue, 17 May 1994 10:32:41 +0200

> > Can't we agree that the user is responsible for this?
>
> Yes, if we acknowledge too that this is the time-honored attitude that
> keeps the masses scared to death of parallel programming <wink>.
>
> [goes on to make many good points]
>
> i.e.-i'd-be-a-lot-happier-with-"na-i-don't-want-to"-than-with-"na-i-
> *shouldn't*-want-to"-ly y'rs - tim

I guess MY point is that to make the entire Python library thread-safe
would be a massive task. It may be necessary to do so at some point,
or at least to document meticulously how safe things are with respect
to threads (like in the Solaris 2 man pages). I don't think it is
very useful to start fixing random things -- either it should be done
properly or no time should be wasted on it. Right at the moment I
don't have time to waste :-)

BTW I do think that the Python library is in a much better position
with regards to thread safeness than the C library (there are very few
routines that use private static work areas and the interpreter has a
basic mutex that serializes most atomic operations). Maybe someone
can make a quick study of what would be required?

--Guido van Rossum, CWI, Amsterdam <Guido.van.Rossum@cwi.nl>
URL: <http://www.cwi.nl/cwi/people/Guido.van.Rossum.html>