Re: GWorkspace: hang on exit. Clang?

[David Chisnall]

On 24 Oct 2012, at 00:50, Riccardo Mottola wrote:

> 2012-10-23 23:22:17.898 GWorkspace[3318] reported exception - <NSException: 
> 0x2bd04344> NAME:NSInvalidReceivePortException REASON:invalidated while 
> awaiting reply INFO:(null)

I've seen this in the NSConnection tests before, but whenever I find a bit of 
time to debug it, the failure stops appearing.  It's quite frustrating: it 
failed 5 times in a row when I didn't have time for debugging, then when I did 
I ran it a hundred times without managing to reproduce the failure.

Since the failure appears and disappears with the same binary, I am inclined to 
suspect a race condition somewhere in the NSConnection code, which various 
compilation modes make more or less likely to be a problem by subtly altering 
the timing of the test.  Whether this race is intrinsic to the source code or 
introduced by a buggy optimisation, I don't know.

David

-- Send from my Jacquard Loom