--- Log opened Tue Nov 21 00:00:32 2023
00:04 -!- lightningbot [lightningb@azure.erisian.com.au] has quit [Remote host closed the connection]
00:04 -!- lightningbot [lightningb@azure.erisian.com.au] has joined #secp256k1
01:33 -!- tromp [~textual@92-110-219-57.cable.dynamic.v4.ziggo.nl] has quit [Quit: My iMac has gone to sleep. ZZZzzz…]
02:50 -!- tromp [~textual@92-110-219-57.cable.dynamic.v4.ziggo.nl] has joined #secp256k1
03:48 -!- RubenSomsen [sid301948@user/rubensomsen] has quit [Ping timeout: 246 seconds]
03:55 -!- RubenSomsen [sid301948@user/rubensomsen] has joined #secp256k1
05:26 < real_or_random> we #define this to _int128  on MSVC, and this one returns a signed _int64
05:28 < real_or_random> roconnor: so I believe we don't want this... unless we make the macro a functional-style macro that casts to uint64_t on MSVC 
05:28 < real_or_random> that sounds minimally cleaner than what we have currently?! i don't know
05:30 < roconnor> Probably best to keep it as is.  I'm not sure why I changed it locally.
05:30 < roconnor> Maybe I have more warnings turned on.
05:31 < real_or_random> makes sense
05:32 < real_or_random> "don't change it if the new version is not clearly better" 
06:19 -!- tromp [~textual@92-110-219-57.cable.dynamic.v4.ziggo.nl] has quit [Ping timeout: 256 seconds]
06:42 -!- preimage [~halosghos@user/halosghost] has joined #secp256k1
08:08 < roconnor> real_or_random: okay I think I see what was bothering me.
08:08 < roconnor> https://github.com/bitcoin-core/secp256k1/blob/e72103932d5421f3ae501f4eba5452b1b454cb6e/src/int128_struct_impl.h#L49C22-L49C22 on line 49
08:08 < roconnor> there is an implicit cast from unsigned to signed.
08:09 < real_or_random> ah I think we have an assumption in assumptions.h for these casts
08:10 < roconnor> yeah, I always think that is UB, but I guess it is implementation defined.
08:13 < roconnor> I think I spent a while trying to figure out an expression that would do the conversion without overflow.
08:14 < real_or_random> hehe yeah,
08:15 < real_or_random> my thinking is that it is just philosophically cleaner if the signed  x signed function also returns signed
08:15 < roconnor> no this is the low bits of the product.  It is very unsigned.
08:15 < real_or_random> the next C standard will set two's complement in stone, by the way
08:15 < roconnor> the thing is that _mul128 is wrong.
08:16 < real_or_random> oh hm
08:17 < real_or_random> in that case, I think I can accept the fact that we're philosophically wrong :P
08:18 < real_or_random> microsoft +shrug+
13:43 -!- preimage [~halosghos@user/halosghost] has quit [Quit: WeeChat 4.1.1]
22:22 -!- bob_x1 [~bob_x@user/bob-x1/x-8934932] has quit [Remote host closed the connection]
22:22 -!- bob_x1 [~bob_x@user/bob-x1/x-8934932] has joined #secp256k1
23:46 -!- tromp [~textual@92-110-219-57.cable.dynamic.v4.ziggo.nl] has joined #secp256k1
--- Log closed Wed Nov 22 00:00:33 2023