***sneek_ is now known as sneek
***catonano_ is now known as catonano
***sneek_ is now known as sneek
***karlosz_ is now known as karlosz
***apteryx_ is now known as apteryx
***jao is now known as Guest86932
<lloda>at this point since all of Guile compiles fine with STRICTNESS set to 2, i'd say the type checking of all user programs is worth some unnamed arch with a bad compiler possibly maybe having slower perf, so we should make STRICTNESS 2 the default. Plus it makes Guile modern C++ compatible. <lloda>Re: [invalidate-compilation-cache is deprecated; recompile your modules] deprecation warning, do I have to recompile my modules manually now? Only sometimes? Autocompilation is on fwiw. <dsmith-work>lloda: I'm curious, have you seen any perf changes with strict=2? <wingo>lloda: most people will not see that error, if they build either with guile 3.0.2 or 3.0.3 <lloda>I was just passing -DSCM_DEBUG_TYPING_STRICTNESS=2 when compiling my code and I didn't notice anything, but I wasn't running benchmarks <lloda>but if I change SCM_DEBUG_TYPING_STRICTNESS to 2 in scm.h, I do see some errors when compiling Guile <wingo>error only appears either for people who define languages, or people that do "git pull" and keep old .go files <lloda>wingo: I deleted all my cache and it appeared again so I dunno <lloda>but I do rebuild Guile itself a lot <lloda>i'm satisfied if I don't have to do anything by hand in a normal situation :-) <lloda>is it right that SCM_NEWSMOB macros in libguile/smob.h cast to (scm_t_bits) directly instead of using PACK/UNPACK? <lloda>ok i've done a little benchmark with 2 vs 1 and I don't see anything meaningful <lloda>dsmith-work: probably deserves more extensive benchmarking tho <lloda>this fixes building Guile with #define SCM_DEBUG_TYPING_STRICTNESS 2 in scm.h <lloda>dsmith-work: yes I saw that when the stricter conversion flagged my == <lloda>it might be a common error, without the check ***tazjin is now known as benry
***benry is now known as tazjin
***sputny1 is now known as sputny