IRC channel logs
2025-08-28.log
back to list of logs
<dthompson>fnat: I really ought to release 0.4.0 soon. it has a few good things in it but probably 0.3.0 will be fine to get started with <fnat>dthompson: Brill, thanks! <cow_2001>oh, i looked into string-contains and i don't know any reasonable amount of c to make these changes :< <cow_2001>i've heard TUPE is good for general UNIXy C <olivermead>How fast is (compiled) Guile, these days? I saw a presentation from 2016 on v2.2 that showed pretty middling performance and he mentioned that v3 would be generating real ELF objects — is this the case? <identity>olivermead: guile indeed uses the elf format for its bytecode, but not in the ‘native binaries’ sense, in the ‘use ELF for object storage because ELF is a good format’, see (info "(guile) Object File Format"). guile has decent performance, not Chez, but in the ballpark. there is a simple template JIT and stuff, but no ‘proper’ native compilation, see (info "(guile) Extending the Compiler") <olivermead>identity: Thank you, should be perfectly good for my needs. <ArneBab>wingo: I now asked four times for merge permission on codeberg guile/guile, without reaction. We have extremely simple PRs on codeberg that do not get merged because no one with permission is active. That sounds extremely unhealthy: we made our review bottleneck *much* worse. So I currently see the codeberg migration as a net negative. If people with push permission on savannah don’t have push permission on codeberg, that migration harms <ArneBab>the one aspect that’s blocking us the most at the moment. <ArneBab>⇒ what are we doing? This is broken. <dsmith>ArneBab, How about poking civodul too? <old>is port-with-print-state still supported? <old>anybody knows if guile can optimize multiple access to a parameter to a single one? Basically doing loads fusing <old>I think it does not, but that could probably be a nice thing to have