--- Day changed Sat Aug 08 2015
14:00 <@andytoshi> for jacobian calculation i have to do a bunch of operation (add/sub/mul/etc) treating the field elements as integers .. should i create a separate uint256_t type for this?
14:01 <@andytoshi> i've been writing code in field_XxY_impl.h and i've been adding a bunch of oddball functions which treat the field elements as integers and i don't feel like they belong there
14:02 <@sipa> do you never need anything above 256 bits in size?
14:02 <@sipa> i think you should add your code as a secp256k1_num_t instance
14:02 <@sipa> (for which only GMP exists)
14:02 <@andytoshi> ah, i just noticed that
14:02 <@andytoshi> yeah, i should do that
14:02 <@andytoshi> well, i never need more than 256 bits
14:03 <@andytoshi> and i don't need signedness
14:03 <@andytoshi> but i guess neither would hurt
14:03 <@sipa> i think none of the current num code needs more
14:03 <@andytoshi> err, supporting larger sizes would hurt, it'd be a ton of extra complexity. signedness doesn't matter
14:04 <@sipa> num_t used to be the backend for scalar operations too
14:04 <@andytoshi> it looks like the num.h API is perfectly sensible for fixed-size uints
14:04 <@andytoshi> ints
14:05 <@sipa> cool
14:06 <@sipa> (the num code is effectively already used for fast modular inverse, it would be hot-pluggable if you swap another implementation im)
14:08 <@andytoshi> yeah, i see, configure.ac sets USE_FIELD_INV_NUM and USE_SCALAR_INV_NUM (which passthrough to the num implementation for field and scalar inverse) if the gmp num implementation is used
14:09 <@andytoshi> cool, so i guess i want a single PR that adds jacobian and field/scalar inverse, and replaces NUM_NONE with NUM_BUILTIN or whatever we wanna call it
14:15 <@sipa> exactlt
14:41 <@andytoshi> sipa: there are two lines in hash_impl.h `#undef ReadBE32` and `#undef WriteBE32`. these symbols are not referred to anywhere else. can you confirm that these are accidental and i can remove them?
14:41 <@andytoshi> (i'm moving the BE32 macro to util.h and adding a BE64 one, is how i noticed this..)
15:11 <@sipa> once i find my glasses i'll have a look
18:48 -!- btcdrak [uid52049@gateway/web/irccloud.com/x-iwwjhxpqngellcow] has quit [Quit: Connection closed for inactivity]
22:22 -!- btcdrak [uid52049@gateway/web/irccloud.com/x-rncrxeixxsfnswkw] has joined #secp256k1