global interpreter lock
fakeaddress at nowhere.org
Fri Sep 2 06:47:07 CEST 2005
Mike Meyer wrote:
> Bryan Olson writes:
>>With Python threads/queues how do I wait for two queues (or
>>locks or semaphores) at one call? (I know some methods to
>>accomplish the same effect, but they suck.)
> By "not as good as", I meant the model they provide isn't as managable
> as the one provided by Queue/Threading. Like async I/O,
> Queue/Threading provides a better model at the cost of
I can't tell why you think that.
> Instead of making vague assertions, why don't you provide us
> with facts?
Yeah, I'll keep doing that. You're the one proclaiming a 'model'
to be more manageable with no evidence.
> I.e. - what are the things you think are obvious that turned
> out not to be true? Name some software that implements sophisticated
> services that we can go look out. And so on...
Thought we went over that. Look at the popular relational-
database engines. Implementing such a service with one line of
execution and async I/O is theoretically possible, but I've not
heard of anyone who has managed to do it. MySQL, PostgreSQL,
IBPhoenix, MaxDB, all have multiple simultaneous threads and/or
processes (as do the competitive commercial database engines,
though you can't look under the hood so easily).
More information about the Python-list