Re: [9front] libmp and error handling

[ori@eigenstate.org]

>> This is still likely to require quite a bit of change in libmp; if
>> it's currently calling sysfatal, then it's very likely full of
>> things like:
>> 
>> 	x = a();
>> 	y = b()
>> 	return mpadd(x, y);
>> 
>> which means that if b() errors, then x is leaked.
> 
> So, I tried changing libmp itself.  It rapidly became a hairy mess of
> return checking, and making the library do more allocation just to
> preserve the original function arguments.  Instead, here's a patch to
> fix the manpage:

Thanks, committed.