Threading blocking functions

[Duncan Booth]

piet at cs.uu.nl wrote in <usnryubiy.fsf at cs.uu.nl>:
>>>>> Roy Katz <katz at Glue.umd.edu> (RK) writes:
>RK> If a function is called simulatenously by different threads,
>RK> will each thread blocand wait for the function to finish executing,
>RK> or will a new copy of the function begin executing? How can I specify
>RK> either behaviour (blocking, nonblocking)?
>
>They may execute simultaneously. If you don't want that you have to set a
>lock.
>
>

It may be of use to know that Zope includes a C extension module called 
'Sync' that provides a mixin class 'Synchronized'. You can use this module 
in non-Zope python programs quite easily (just make sure that Sync.pyd is 
somewhere on your PythonPath and then import Sync).

Basically, any class that includes Sync.Synchronized as a base class will 
automatically single thread calls to all of its member functions. So the 
original example:

RK> def update( data ):
RK>    do-something-with-data

could be implemented as:

import Sync
class updateclass(Sync.Synchronized):
   def update(self, data):
        do-something-with-data
update = updateclass().update

Every instance creates its own lock and stores it in the _sync__lock member 
of the class.