From pablo.yabo at gmail.com  Sat Nov  3 22:58:06 2007
From: pablo.yabo at gmail.com (Pablo Yabo)
Date: Sat, 3 Nov 2007 18:58:06 -0300
Subject: [capi-sig] Thread structures BUG?
Message-ID: <db6698600711031458m49e2ec32uca4f114fe9be232d@mail.gmail.com>

Hello,
I create lots of threads that run a script and I have a problem when a
thread is created with the same threadId that existed before.

When a new thread is started I use this code to enter the interpreter:
    // create a thread state object for this thread
PyEval_AcquireLock();
threadState = PyThreadState_New(mainInterpreterState);
PyThreadState_Swap(threadState);

and when the thread finishes:

PyThreadState_Swap(NULL);
PyThreadState_Clear(info);
// delete my thread state object
PyThreadState_Delete(info);
PyEval_ReleaseLock();


Everything works smoothly until a thread id is re-used.
I've been testing what is happening and the problem is that
PyGILState_GetThisThreadState() returns a PyThreadState for the reused
thread id.
I changed my code to call PyGILState_GetThisThreadState() before creating
the PyThreadState. But if I try to use the returned PyThreadState (that
shouldn't exist because I've already deleted) I got an error when trying to
access the interpreter dictionary.

The workaround that I found is to keep all PyThreadState structures without
calling PyThreadState_Clear / PyThreadState_Delete. When a new thread is
created I call PyGILState_GetThisThreadState(). If it returns something I
reuse the structure.

In this way, my code works.

Thanks,
Pablo Yabo

--
http://www.nektra.com

From sndive at gmail.com  Fri Nov  9 20:04:01 2007
From: sndive at gmail.com (Anton Tropashko)
Date: Fri, 9 Nov 2007 13:04:01 -0600
Subject: [capi-sig] No pickling?
Message-ID: <2822e000711091104h16258b0bl962fb068d387eb8d@mail.gmail.com>

I see some api published for marshal module:
http://www.python.org/dev/doc/maint24/api/marshalling-utils.html
but per latest discussion on comp.lang.python
people were suggesting to use pickle.

If i pass the marshalled string to a machine running python on
a machine with a different byte order, would it unmarshall correctly?

From fdrake at gmail.com  Fri Nov  9 20:20:34 2007
From: fdrake at gmail.com (Fred Drake)
Date: Fri, 9 Nov 2007 14:20:34 -0500
Subject: [capi-sig] No pickling?
In-Reply-To: <2822e000711091104h16258b0bl962fb068d387eb8d@mail.gmail.com>
References: <2822e000711091104h16258b0bl962fb068d387eb8d@mail.gmail.com>
Message-ID: <9cee7ab80711091120p221c96c4j83a93e419f39a563@mail.gmail.com>

On Nov 9, 2007 2:04 PM, Anton Tropashko <sndive at gmail.com> wrote:
> If i pass the marshalled string to a machine running python on
> a machine with a different byte order, would it unmarshall correctly?

This isn't the right place for this question; it makes more sense on
comp.lang.python.

Both the marshal and pickle formats are byte-order independent.


  -Fred

-- 
Fred L. Drake, Jr.    <fdrake at gmail.com>
"Chaos is the score upon which reality is written." --Henry Miller

From sndive at gmail.com  Fri Nov  9 22:55:32 2007
From: sndive at gmail.com (Anton Tropashko)
Date: Fri, 9 Nov 2007 15:55:32 -0600
Subject: [capi-sig] scopes and PyEval_EvalCode()
Message-ID: <2822e000711091355x1a1f094dl223d910c22b98bda@mail.gmail.com>

On Nov 6, 8:25 pm, snd... at gmail.com wrote:
> i naively created execution context:
>     PyObject *execcontext = PyDict_New();
> stuffed a handle in it:
>     PyObject *ih = PyCObject_FromVoidPtr(handle, NULL);
>     int st= PyDict_SetItemString(res, "interp", ih);
>
> and later on in a function for a module that i defined
> expected to extract that handle:
>         PyObject *dict = PyEval_GetGlobals(); // or GetLocals
> depending on where
> // execcontext ended up in PyEval_EvalCode
>         PyObject *co = PyDict_GetItemString(dict, "interp");
>         assert(PyCObject_Check(co));
> but i get null co either way and the dict does not match execcontext
> that is passed into PyEval_EvalCode.
>
> any ideas how to fix this?

in other words: is there a scope more global than a module scope?
something that won't change for a given PyEval_EvalCode or a similar call.
i stuffed my CObject into __main__ but in case of multiple threads or
nested PyEval_EvalCode i'd have a problem with this approach.

i somehow just need to get to the pointer to the same dictionary that
i passed to PyEval_EvalCode for global scope to retrieve my handle
in a method of a new python type defined in C space.

thank you