On Wed, Dec 09, 2009 at 03:18:27PM +0100, Thomas Jansen wrote: > On Tue, Dec 08, 2009 at 08:22:09PM +0100, Diego Biurrun wrote: > > On Sun, Nov 29, 2009 at 11:45:26AM +0100, Tobias Heer wrote: > > > ./libpisa/debug.c:522:void pisa_uint8_to_binstring(uint8_t val, char > > > *buffer) > > > ./libpisa/debug.h:150:void pisa_uint8_to_binstring(uint8_t val, char > > > *buffer); > > > ./libpisa/debug.c:543:void pisa_uint16_to_binstring(uint16_t val, char > > > *buffer) > > > ./libpisa/debug.h:151:void pisa_uint16_to_binstring(uint16_t val, char > > > *buffer); > > > ./libpisa/debug.c:564:void pisa_uint32_to_binstring(uint32_t val, char > > > *buffer) > > > ./libpisa/debug.h:152:void pisa_uint32_to_binstring(uint32_t val, char > > > *buffer); > > > > I think these can go away, Thomas? > > This is rather specialized debug code to get a binary representation of a > uintXY_t. I'd say it's not needed. If you really need to know what the value > of a certain bit is, you can easily write some printf("%s", variable&mask ? : > "true" : "false); code. Or you could print the whole variable and use a > calculator. Even the hexdump used above could be used. > > Verdict: Remove. Removed, everything else kept. I also removed a few unused inline functions and moved some that only got used in one place to the corresponding .c file. > > > ./libpisa/packet.h:137:uint8_t pisa_get_packet_version(const pisa_packet > > > *pkt); > > > ./libpisa/packet.c:131:uint8_t pisa_get_packet_version(const pisa_packet > > > *pkt) > > > ./libpisa/packet.h:140:pisa_packet_flags pisa_get_packet_flags(const > > > pisa_packet *pkt); > > > ./libpisa/packet.c:164:pisa_packet_flags pisa_get_packet_flags(const > > > pisa_packet *pkt) > > > ./libpisa/packet.h:145:void pisa_set_packet_flags(pisa_packet *pkt, > > > pisa_packet_flags flags); > > > ./libpisa/packet.c:208:void pisa_set_packet_flags(pisa_packet *pkt, > > > pisa_packet_flags flags) > > > > There are similar functions for other struct members, but the functions > > are very trivial. Keep? Remove? Thomas? > > Obviously, the uint8_t fields of pisa_packet don't need byteorder conversion, > rendering the accessor functions useless. The reason to keep them might be the > consistent API for manipulating pisa_packet structures. Also note, that > pisa_get_packet_version is used in packet.h in the inline function at the > bottom of the file. So I'd rather keep them. > > Verdict: Keep. > > However I see another interesting question in that context: What exactly is > the > purpose of the flags field in the structure? Right now we don't have any flags > defined and I don't see that coming in the near future. Might be an option to > throw out the flags from the structure (and because of the packet format > change > bump the version number). Then we could get rid of the two accessor functions > for flags and keep the get version one, which is a false positive anyway. What do you mean by false positive? Otherwise this sounds like a good idea, implement it :) Diego