[Stackless] Strange crashes after channel send/receive

[Lloyd Weehuizen]

Hi Chris

Christian Tismer wrote:
> 
>> Python is embedded in my application and the main update loop calls 
> 
> Besides the fact that doing things this way guarantees worst
> possible performance :-) - (I assume that you are calling into
> uninitialized Stackless all the time, and the main tasklet
> gets re-created all the time?) -
> are you absolutely sure that you don't call into things
> in your C code that is non-reentrant? This is easy to do
> with Stackless.

Wow, ok, thanks for that tip, I was planning on moving the scheduling 
completely into C++ to bypass the python update at a later time, 
although I didn't realise it was creating a whole new main tasklet each 
time currently no.

>> As far as I can tell it gets triggered after the C++ app sends a 
> 
> Please add checks about what your real main tasklet is.
> Maybe you are referring to a zombie?

The problem seems to have gone, I rolled back the code and placed a lot 
more asserts throughout the python interface layers and did find a few 
nasty issues to do with PyObjects, however I can't pin point exactly 
which one was breaking things. As you say it was probably worsened by 
reentrant code.

Its definitely not a stackless problem, just my lack of experience with 
the Python API :)

> 
>> Does anybody have any tips on the best way to start debugging a 
> 
> There is not too much. I recommend a debug build, of course.
> And you can register callbacks on schedule() calls and channel
> operations for debugging.

Thanks, will take a look at the schedule callbacks...

> 
> If you don't mind, you also can send me the whole project
> and I can take a look.

Unfortunately I'm not able to send the code out at this time :(

> cheers - chris

Thanks,
Lloyd

_______________________________________________
Stackless mailing list
Stackless at stackless.com
http://www.stackless.com/mailman/listinfo/stackless