Experimental Physics and
Industrial Control System

<[email protected]> · Mon, 22 Apr 2013 07:36:02 +0000

From: [email protected] [mailto:tech-talk-
> On 2013-04-19 [email protected] wrote:
> > I've just discovered that calling ca_context_create(1) after
> > iocInit() appears to succeed (it returns ECA_NORMAL) ... but the
> > resulting context is not very healthy!  In my particular case I find that
> > ca_array_get_callback delivers zero length data, but I expect other deeper
> > problems are to be found.

I have just attempted to reproduce this problem in a standalone fragment of code (the problem was previously occurring in a complex and dynamic Python program) ... and so far have completely failed!  I'll post an update if I succeed in pinning the problem down more closely.

> Is your code running in the main thread, or its own private thread?

This is code all called from the main thread.

> You can use ca_current_context() to find out if there is context set for the
> current thread yet.  This returns a NULL pointer if no context has been
> created yet.  If the result is not NULL you may be able to use the
> undocumented (and mis-spelled) routine (with a prototype in cadef.h)
>     int ca_preemtive_callback_is_enabled(void);
> to find out whether that context is preemptive or not.

This was very helpful (yes, what a horrible typo, it threw me for a while, as it's somehow quite hard to spot).  Turns out I *am* the first create the CA context, and as I've noted above, I can't reproduce the problem in isolation yet ... will carry on digging and will report if I find anything.

Experimental Physics and Industrial Control System

Experimental Physics and
Industrial Control System