IRC channel logs

<ijp> http://shift-reset.com/pastes/mergesort_new_inliner.html

<ijp>much less nesting than before

<ijp>there are some unnecessary aliases, but I will clear that up shortly

***karswell` is now known as karswell

<amz3>héllo :)

<ijp>I don't know how many of you saw it yesterday, but http://shift-reset.com/pastes/mergesort_new_inliner.html is starting to look much less messy

<davexunit>ijp: looking good

<ijp>I can handle the realiasing (var foo = bar;), but I need to change the representation if I'm going to inline the constants

<amz3>I have a error that is kind of cryptic: Unbound variable: ((#{ g33171}#) <node> #f)

<ijp>and how did you obtain this error

<amz3>well, I defined a record with a define-record-type* <node>, it's not srfi-99 i changed to make <node> become node-identifier

<ijp>code, not prose, please

<amz3>huhu

<amz3>ok

<amz3>well sending the whole thing is fat

<amz3>what is #{ g33171}# N

<amz3>?

<ijp>it's a symbol

<amz3>I will make a minimal example code

<amz3>it's compilcated it requires guix

<amz3>I suspect there is funny symbol somewhere

<ijp>probably introduced hygienically

<amz3>paron_remote: is there somewhere I can fetch pamrel.el ?

<amz3> http://pamrel.lu/45324/

<amz3>that said

<amz3>I found something

<amz3>when I do C-c C-k and then C-x C-e on the last sexp it doesn't work

<amz3>then if if I C-x C-e the record and eval the last exp it works

<ijp>hmm, we really need to get a proper srfi-99 setup one of these days. the record system is very similar anyway

<amz3>(my define-record-type* is mainly a hack to have proper highlighting...)

<amz3>(I saw it while reading your code :p)

<amz3>I think that the srfi-99 that you use is ok at least for my usecase

<amz3>(that with set-field and set-fields makes working with record a joy)

<ijp>so, I'm not seeing any buggy behaviour

<amz3>my paste is missing something

<amz3>ijp: do you see (pk (node-identifier ...) at the end ?

<ijp>yes

<amz3>ok let it go I will sort this out

<amz3>thx

<ijp>it could be somethign that is fixed on master, but I can't think what it would be

<amz3>like I said there seems to be a way to make it work, and it has something to do with emacs/geiser

<amz3>it definitly emacs/geiser that doesn something

<remi`bd>hello! Guile’s manual says that `(sizeof TYPE)`, where TYPE is a list, will return the size of the structure that matches TYPE with “ABI conventional packing”

<remi`bd>what is this “ABI conventional packing”? Does that mean `sizeof` takes structures alignment padding into account?

<remi`bd>the reason for my question is: the size of `struct foo { char *bar; unsigned int baz; }` varies between C and Guile

<remi`bd>(sources: http://paste.lisp.org/display/150153 & http://paste.lisp.org/display/150154 — see the annotations for output)

<Feufochmar>Hello. I'm encountering a weird bug with match:substring (I think). Given the data file http://feuforeve.fr/vrac/test-guile.txt and the guile script http://feuforeve.fr/vrac/test-guile.scm , I'm getting an error (value out of range, see http://feuforeve.fr/vrac/error-guile.txt ) when I run guile non-interactively. If I copy/paste the contents of test-guile.scm inside the REPL, I don't encounter any problem.

<mark_weaver>Feufochmar: you need to call (setlocale LC_ALL "") near the beginning of your script.

<mark_weaver>when running guile interactively, that is done automatically.

<mark_weaver>but in a script, you need to do it explicitly or else the C locale is assumed, and so your text file cannot be decoded properly.

<mark_weaver>(in guile-2.2 it will be done automatically for scripts as well, but not in guile-2.0)

<Feufochmar>mark_weaver: guile does not use environment variables like LANG or LC_ALL set in the environment ?

<mark_weaver>Feufochmar: it does if you call (setlocale LC_ALL "")

<mark_weaver>I agree that we should do it automatically for scripts, and we will in guile 2.2, but for backward compatibility reasons we can't change the behavior now in guile-2.0.

<Feufochmar>ok, thanks

<mark_weaver>np

<mark_weaver>Feufochmar: there's also an environment variable you can set to tell it to always set the locale when starting up scripts: GUILE_INSTALL_LOCALE

<mark_weaver>however, it's probably better to call (setlocale LC_ALL "") so that your script will work properly regardless of the environment variable settings.

<mark_weaver>(and in case you run any scripts that depend on the locale being left alone)

<mark_weaver>anyway, you can set that variable to "1" (to install the locale) or "0" (to not install the locale)

<mark_weaver>also, we didn't add support for GUILE_INSTALL_LOCALE until guile-2.0.9

<paron_remote>amz3: yeah

<paron_remote>amz3: https://github.com/cwebber/pamrel-el

<paron_remote>also, hello *

<paron_remote>ijp: woo re: mergesort_new_inliner.html

***michel_ml is now known as michel_mno

***heroux_ is now known as heroux

***ft_ is now known as ft

***heroux_ is now known as heroux

***michel_mno_afk is now known as michel_mno