On Sat, Jun 30, 2012 at 12:34 AM, Rob
<address@hidden> wrote:
> Thanks for that - i'm happy to accept a picky answer :). i have since found
> that i need to use a typecast for this type of case with gcc, anyway, so
> i'm refactoring now.
Why's that the case? Just curious, from what I could tell gcc accepts function
pointers there just fine.
The error message (from gcc) was that we were casting incompatible types. Both had the same signature, but didn't use a typedef to represent that. It seems to me to be a buggy corner case in gcc, but it was easy to please by replacing the duplicated function signature code with a prettier typedef.
Yeah - a typedef is much better than what I proposed.
tcc accepts it fine here.
sqlite4 now compiles just fine with tcc on x64 (and blazing fast - about 8x faster than gcc), but... the test suite crashes on tcc and i don't know why. i hope to take a look at that as the v4 tree stabilizes somewhat (it's very much pre-beta right now).