[asterisk-dev] [patch] Alignment fix for main/event.c

Jason Parker jparker at digium.com
Tue Dec 11 08:54:01 CST 2007


Rafal Boni wrote:
> I whined about this in my earlier email -- if you're running on a
> machine that cares about alignment constraints of data (sparc64 in my
> case), event.c has at least a few gotchas and will likely cause a crash
> before Asterisk can answer the 1st call..
>
> <obcommentary>
> I also am not sure that if I wanted the events to be wire-shippable
> between systems that I'd do it the way it's done there -- especially
> trying to ship enums in any on-wire transport seems pretty dangerous.  A
> a better idea ISTM would be to leave the events unmolested inside the
> system and build a marshalling / unmarshalling layer for putting them on
> the wire.  Then you don't care as much about all this enddian-frobbing,
> alignment and buffer-walking if you're just queueing / passing them
> around inside a single process (or even single machine).
>
> All that from looking at a couple of source files in isolation ;)
> </obcommentary>
>
> Patch which gets my trunk build to actually run the demo config from a
> softphone, and succesfully dial out to digium's test server is attached.
>
> --rafal
>   

These patches will have to be put through the bug tracker at
bugs.digium.com so that they can be tracked properly.



More information about the asterisk-dev mailing list