On Sat, Jun 16, 2012 at 05:29:26PM +0200, Leon Timmermans wrote: > On Sun, Jun 10, 2012 at 7:59 PM, Father Chrysostomos <sprout@cpan.org> wrote: > > I thought increasing the size of an argument was not considered an > > incompatible change. We have already done that with some functions. > > I'd agree. The only place where widening them up can really break is > function pointers, and I doubt anyone is doing that (if only because > of the pTHX mess they'd have to do). Widening the types of pointers breaks things. (That's probably not quite phrased correctly) Jarkko spent quite a lot of time a decade ago fixing bugs that DEC Alphas exposed, where the internals were taking a pointer to (IIRC) a STRLEN variable and passing it to a function expecting a pointer to (IIRC) a U32 or suchlike. Until then, on most platforms people had ever actively developed perl on, U32 and STRLEN had the same size. Nicholas ClarkThread Previous | Thread Next