--- Log opened Sat Aug 28 00:00:51 2021
02:01 -!- jonasschnelli_ is now known as jonasschnelli
06:38 -!- luke-jr [~luke-jr@user/luke-jr] has quit [Ping timeout: 250 seconds]
06:44 < roconnor> https://gcc.gnu.org/onlinedocs/gcc-4.7.2/gcc/Function-Attributes.html#index-g_t_0040code_007bnonnull_007d-function-attribute-2564
06:44 < roconnor> The compiler may also choose to make optimizations based on the knowledge that certain function arguments will not be null. 
06:45 < roconnor> I have some, totally unverified, indications that the ARG_CHECKS that arguments are non-null are being compiled away by the non-null attributes.
06:46 < roconnor> Even if this is really happening, I'm not entirely sure if it is a problem, though I suspect it would be.
06:48 < sipa> just checked
06:48 < sipa> SECP256K1_ARG_NONNULL is defined as nothing when SECP256K1_BUILD is defined
06:48 < roconnor> really?
06:49 < sipa> see include/secp256k1.h, line 169
06:49 < roconnor> oh.
06:49 < roconnor> I did look at that line, but I guess I only glanced.
06:52 < roconnor> Okay that's good.  I encounted this issue in a non-BUILD build.
06:56 < roconnor> Damn it, there is even a comment about this
06:56 < roconnor> * NONNULL is not used if SECP256K1_BUILD is set to avoid the compiler optimizing out
06:58 -!- siv2r09300 [~siv2r@103.77.37.144] has joined #secp256k1
07:16 -!- jonatack [~jonatack@user/jonatack] has joined #secp256k1
08:29 -!- luke-jr [~luke-jr@user/luke-jr] has joined #secp256k1
09:21 -!- lukedashjr [~luke-jr@user/luke-jr] has joined #secp256k1
09:23 -!- luke-jr [~luke-jr@user/luke-jr] has quit [Ping timeout: 252 seconds]
09:24 -!- lukedashjr is now known as luke-jr
09:46 -!- lukedashjr [~luke-jr@user/luke-jr] has joined #secp256k1
09:47 -!- luke-jr [~luke-jr@user/luke-jr] has quit [Ping timeout: 250 seconds]
09:48 -!- lukedashjr is now known as luke-jr
09:55 -!- siv2r09300 [~siv2r@103.77.37.144] has quit [Quit: The Lounge - https://thelounge.chat]
11:24 -!- siv2r09300 [~siv2r@103.77.37.144] has joined #secp256k1
13:11 < gmaxwell> roconnor: yeah, we wanted the behavior of warning the caller if they're calling it wrong without the optimizing out which isn't a useful thing for us.  The code should however still be correct with those checks optimized out, at least for correct callers.
13:12 < gmaxwell> roconnor: it shouldn't be easy to do non-build builds anymore, setting SECP256K1_BUILD is supposed to be automatic now.
14:33 < roconnor> Right.  I think I was working from a branch that was prior to your commit.
14:33 -!- roconnor [~roconnor@host-45-58-210-199.dyn.295.ca] has quit [Quit: Konversation terminated!]
17:15 -!- belcher_ [~belcher@user/belcher] has joined #secp256k1
17:18 -!- belcher [~belcher@user/belcher] has quit [Ping timeout: 248 seconds]
17:37 -!- siv2r09300 [~siv2r@103.77.37.144] has quit [Quit: The Lounge - https://thelounge.chat]
18:32 -!- siv2r09300 [~siv2r@103.77.37.178] has joined #secp256k1
19:57 -!- siv2r09300 [~siv2r@103.77.37.178] has quit [Ping timeout: 240 seconds]
--- Log closed Sun Aug 29 00:00:52 2021