[asterisk-dev] Pthread wrapper updates

[SF Markus Elfring]

> Sure, but it does nothing to improve the situation except to reiterate
> that the code could be better.

I would like to prepare if obstacles on this topic can be removed.


> None of the places that you find that don't check the return values are
> doing it because we make a conscious decision not to; they are doing it
> because that's the pattern that was always present in the code and
> everyone follows the existing pattern.

It seems that the administrator "russell" has got a similar view.
http://bugs.digium.com/view.php?id=8157:
"That issue is not a bug, it is intentional. [...]"

But I try to convince the code maintainers from the opposite. Can anybody other than me see the specific details that are wrong with the current approach?
I'd like to get support and acceptance for code updates that will be required for correctness.

1. Introduction of a general termination function
    http://opengroup.org/onlinepubs/009695399/functions/abort.html

Which parameters would be needed for an "ast_abort"?

2. Use of "ast_abort" (no "DO_THREAD_CRASH") by the wrapper implementations after error detection

3. Change return type to "void" for some interfaces


> If you want to improve the situation, supply patches that do so...

I want to achieve a consens on the proposed corrections.

Regards,
Markus