On 18 Nov 2005 at 21:29, Joerg Wunsch wrote: > Dmitry Xmelkov provided speed-improved versions of itoa() & Co. that > no longer use standard division functions for formatting: > > https://savannah.nongnu.org/patch/?func=detailitem&item_id=3729 > > The only drawback of that approach is that its use will restrict the > posible base of these functions to either 2, 8, 10, or 16. > > I tend to see this as benign, as I cannot imagine why anyone would > want to print out numbers base 7 or base 13, so I'm inclined to > replace the existing itoa() family by Dmitry's submission. When doing > so, I'll probably rename the existing implementations to itoa_full() > etc. to preserve them in case anyone really needs that functionality. > > While itoa() is not really standardized in any way (at least not for > C, I'm not sure for C++), it is commonly implemented with a possible > base of [2..36], so changing that would constitute an API change. > > Opinions? >
I think it would be a good move. In most C libraries the itoa function is extremely slow because it caters for all bases from 2 to 36. I think having a library compile option that compiles the full version in stead of the faster version would be better than renaming it. Default build uses the fast routines, and if someone wants the full version, they can recompile with that option selected. Regards Anton Erasmus-- A J Erasmus _______________________________________________ AVR-GCC-list mailing list [email protected] http://lists.nongnu.org/mailman/listinfo/avr-gcc-list
