On Sun, 2003-07-06 at 23:47, Henrik Nordstrom wrote: > Robert: Please verify that this is correct. This is mostly magic to me > and I do not understand why the linker would ever require a delete > operator if you never create or delete any objects manually even if > these are declared in the class.
Right. Well, it's correct. The glue was there before I generalised the cbdata facility out to List.h. So, removing the glue is safe. CBDATA is needed for the data handed to comm_add_close_handler. Now, as for the optimiser removing the need for a method that optimisation, I'm not sure. Given that we don't allocate DeferredReads via new or delete, it would need some study of the preprocessor output + grepping the output of objdump -tC *.o for DefeferredRead::operator new to identify what was triggering it. Cheers, Rob -- GPG key available at: <http://members.aardvark.net.au/lifeless/keys.txt>.
signature.asc
Description: This is a digitally signed message part
