IRC channel logs

2020-05-08.log

back to list of logs

<unknown_lamer>grep RunHooks -r source/ is your best bet tbh
<unknown_lamer>looking at this it's kind of arbitrary depending on the hook
<unknown_lamer>wow, I never got rid of the pre-STL String class entirely either
<unknown_lamer>maybe I can try dusting this off to remember some C++ and then figure out http://coccinelle.lip6.fr/ to get rid of String
<unknown_lamer>The following packages will be REMOVED:
<unknown_lamer> guile-1.8
<unknown_lamer>I have apparently not updated guile in a *bit*
<unknown_lamer>dsmith: hrm, I applied the patch but after building I get a lot of "Utils.C:(.text+0x1bf3): undefined reference to `scm_from_locale_string'" and whatnot, run into that at all?
<dsmith>No.
<dsmith>Do you have more than one guile installed?
<dsmith>Prob need to remove all -dev packages too.
<unknown_lamer>just guile-3.0-dev guile-3.0-doc guile-3.0-libs guile-3.0 and it's trying to link against libguile-3.0
<unknown_lamer>hrm hrm
<unknown_lamer>will have to take another look later
<dsmith> Nope. Just did a fresh build from my branch. No build errors. Some warnings about "throw".
<dsmith>Hmm. I've got guile 3.0.1 installed here.
<dsmith>My g++ is g++ (Debian 8.3.0-6) 8.3.0
<unknown_lamer>I think it's because I have guile.m4 from 2008 manually included
<unknown_lamer>the idea being that you can reconf without needing guile, in case someone wanted a scriptless bot
<dsmith>"git clean -dxf; autoreconf --install --symlink && ./configure && make"
<unknown_lamer>dsmith: yeah it wasn't that
<unknown_lamer>dsmith: which version of autotools are you using? wondering if maybe i'm running into some incompat there now... (keeps whining about missing config.rpath too)
<unknown_lamer>I am ... so out of date now :-\
<daviid>unknown_lamer: guile.m4 has been patched a couple of times since, I would grab the latest and use it to replace the one you keep in yur project
<unknown_lamer>alas, no dice
<unknown_lamer>so weird
<dsmith>autoconf 2.69-11
<dsmith>nope
<dsmith>yep
<dsmith>automake 1:1.16.1-4
<ahmed1hsn>How do guile compare with Chicken scheme?
<dsmith>(basically, up to date Debian Buster)
<unknown_lamer>dsmith: hrm even weirder, same here (on sid though)
<unknown_lamer>turns out it's something with gcc 9
<dsmith>sneek: version?
<sneek>Sneeky bot running on Guile version 3.0.0 using bobot++ 2.3.0-darcs
***apteryx is now known as Guest91892
***apteryx_ is now known as apteryx
<reepca>is there a way in guile-fibers to properly spawn a new thread that could be considered "outside" of fibers? Closest I've found so far is: (parameterize (((@@ (fibers internal) current-fiber) #f)) (call-with-new-thread ...))
<reepca>for context, I'm trying to implement a thread pool for various pesky blocking operations
<reepca>it works fine to use channels to communicate between "outside fibers" threads and "inside fibers" threads, but if you try spawning a new thread from inside fibers it can't use channels to communicate with anything because it tries suspending through the call-with-new-thread continuation barrier.
<dsmith-work>Happy Friday, Guilers!!
<mwette>Howdy!
<dsmith-work>sneek: botsnack
<sneek>:)
<dsmith-work>sneek: seen unknown_lamer ?
<sneek>unknown_lamer was last seen in #guile 13 hours ago, saying: turns out it's something with gcc 9.
<dsmith-work>Heya mwette
<mwette>testing the bot?
<dsmith-work>Yeah. New bobot++ code.
<dsmith-work>sneek: version
<sneek>Sneeky bot running on Guile version 3.0.0 using bobot++ 2.3.0-darcs
<dsmith-work>Hmm. That version really should be more accurate
<mwette>sneek: later tell dsmith-work nice work on the bot
<sneek>Got it.
<dsmith-work>thanks
<sneek>dsmith-work, you have 1 message!
<sneek>dsmith-work, mwette says: nice work on the bot
***jonsger1 is now known as jonsger
***jao is now known as Guest87292
<stis>hey guilers!
***Aurora_iz_kosmos is now known as Aurora_v_kosmose
<dsmith-work>Bah! Snowing outside!
<manumanumanu>civodul, wingo : If I were to submit a patch with a "simplify-equal-and-eqv" in the style of maybe-simplify-to-eq of primitives.scm, would you be interested? As I wrote to ludo, porting aleix's json parser from match (using equal?) to regular cond clauses led to a whopping 25% speedup on some already pretty fast code (we got the json parser down from a rather slow 19s for a 24mb file, to about 2.5 s on a 2015
<manumanumanu>imac). That rhymes badly with match expanding to optimal code. Since guile is already optimizing equal? to eq? for nulls and booleans, why not eqv? it for bignums, floats (????? I don't know if that is sane), ratnums and chars?
<manumanumanu>The maybe-simplify-to-eq is simple enough (but not as wide as it could be, since keywords are also eq?, but are not in the code). I can use that as a starting point.
<wingo>manumanumanu: plz send the patch :)
<manumanumanu>wingo: the question is then: how can I hack on it without having to do recompiles? Can I chaange the file on a compiled system? (I have never really dug this deep before. I am in waay over my head) Point me in the right direction and I won't bother you again!
<wingo>manumanumanu: do you work from a git checkout?
<manumanumanu>I could. I never got guile to compile on my mac outside of homebrew, but I have a bunch of SBC's that run different versions of guile (2.9.3 - 3.0.1) from git.
<manumanumanu>wingo: sorry, doing my going to bed routine here. answer latency is variable
<manumanumanu>I'll see whatever answer tomorrow! :D Good night! Compiling the latest master on my rock pi 4. It might be done in the morning...
<wingo>civodul: i wired up the baseline compiler to -O0. probably it's broken for many cases but you can give it a go now and fix its bugs ;)
<civodul>wingo: woohoo, a big thank you!
<civodul>i'll give it a spin over the week-end
<wingo>coolio
<wingo>civodul: note i also added an #:optimization-level kwarg to the compile procedures
<rekado_>ooh, nice!
<wingo>if you don't use that, then instead pass #:cps? #f in the opts
<wingo>but better to pass #:optimization-level or -O0 on command line
<wingo>anyway, zzz
<wingo>night :)
<civodul>night!