Re: GNUstep Make not adding -fblocks

[David Chisnall]

On 16 Feb 2011, at 14:57, Richard Frith-Macdonald wrote:

> I can't see that in my email archive ... mybe I accidentally deleted it, or 
> perhaps it was a private email to Nicola.

I thought I sent a mail to Nicola, but looking at my mail history I actually 
just made a note to myself to mail Nicola when I'd done some testing, which 
would explain why neither you nor he received it.

>> It seems that the configure script is testing whether exceptions work, then 
>> deciding that, even though they do, it won't let you use them because it 
>> doesn't like the version of GCC that you have.
> 
> That may well be the case ... because we don't just need to check that we can 
> throw/catch exceptions, we also need to check that we can set a handler for 
> uncaught exceptions, and it may well have been a lot easier to just test for 
> the version of gcc which added support for that to the runtime than to check 
> to see what really happens.
> 
> So, a good test for exception support would check both that the compiler 
> generates exception code and that the runtime supports setting a handler for 
> uncaught exceptions (and that both actually work).
> I seem to remember hacking some sort of test into the base config scripts, 
> but I don't recall whether it was actually what I'd call a good one.

The test seems to be rejecting GCC 4.2: I've not had any problems with native 
exceptions in any of the 4.x series - including the pop-up when apps throw an 
exception that isn't caught - and I think this was also working fine in the 3.x 
series when I was using that, but it's been a few years...

We also check for the availability of the uncaught exception handler in a 
separate test, so it would make sense to just disable native exceptions if this 
doesn't work.

David