Re: Changesets: Re: Parallel access to data created with Octave

[Jaroslav Hajek]

On Tue, May 25, 2010 at 10:55 AM, Michael D. Godfrey
<address@hidden> wrote:
> On 5/25/10 8:43 AM, Jaroslav Hajek wrote:
>>
>> Although parts of this work are useful per se, it feels more like a
>> workaround than a solution. What you really need is a thread-safe
>> reference-counting.
>>
>
> How hard is this to implement?  I seems to me that there will be increasing
> demand
> for thread-safe operation and work-arounds are not the way to get there.
>
> Michael
>
>

In the simplest setup, counter increments and decrements should be
decorated by #pragma omp atomic.
I think we could have a set of macros, similar to Py_INCREF et al, for
doing this.

I do, however, think, that in order to make Octave multithreaded, we
would eventually need to implement some kind of global interpreter
lock, similar to what Python has. There are also open questions what
to do with things like the call stack or symbol table.

-- 
RNDr. Jaroslav Hajek, PhD
computing expert & GNU Octave developer
Aeronautical Research and Test Institute (VZLU)
Prague, Czech Republic
url: www.highegg.matfyz.cz