IRC channel logs

<paroneayea>my package is failing to install modules with make install and I can't figure out why

<daviid>I'm writing a latex.m4 macros that uses guile, that depends on a tiny module that I intend to place in buid-aux/latex.scm

<daviid>so the main macro looks like this (pasting 4 lines)

<daviid>AC_DEFUN([LATEX_CHECK],

<daviid> [#AC_REQUIRE([LATEX_GUILE_M4_SUPPORT_CHECK])

<daviid> $GUILE -c "(add-to-load-path $(top_builddir)) (use-modules (build-aux latex)) $2" > /dev/null 2>&1

<daviid> $1=$?

<daviid> ])

<daviid>though the above does not work of course, that is my question:

<daviid>what should I do to have this subsituted: (add-to-load-path $(top_builddir)), anyone ?

<mwette>$GUILE -L "$(top_builddir)" -c "(use-modules (build-aux latex)) $2" ???

<daviid>mwette: ah, let me try, thanks a lot for your help

<daviid>better, but still fails

<daviid>checking for \\documentclass{standalone}... ./configure: line 12111: top_builddir: command not found

<daviid>no

<daviid>probably some stupid tipo on my side, let me double check

<daviid>the substitution leads to this line

<daviid>$GUILE -L $(top_builddir) -c "(use-modules (build-aux latex)) ... so it miss the "

<daviid>hum, added, but now i see there is no substitution, here is the line after my last change:

<daviid>$GUILE -L "$(top_builddir)" -c "(use-modules (build-aux latex)) (exit ((lambda () (latex-check-for-document-class \\"standalone\\"))))" > /dev/null 2>&1

<daviid>gnulib relocatable.m4 does this: ... "\\$(top_builddir)" ... but here it does not trigger substitution, in the configure file I get "\\$(top_builddir)"

<daviid>gnulib gettext.m4 does almost the same, but: ... "\\${top_builddir}" ... no substitution either, so in configure i get this line

<daviid>very frustrating :(

<daviid>#autotools did not answer so far either, they probably all know ...

<rekado_>I’d like to edit some XML with Guile. There is apply-templates, but it requires unambiguous match descriptions for nodes.

<rekado_>is it possible to use sxpath to describe nodes instead?

<rekado_>that would be the prettiest interface: zoom to a node via sxpath and return a new tree.

<janneke>rekado_: i've used xml->sxml once, really great

<rekado_>janneke: oh, yes, I’m working on sxml, actually

<rekado_>I’d like to be able to write a few xpath specs and then ask for a new tree, where each match has been replaced with the result of a handler.

<wingo>is daniel hartwig still around?

<rekado_>the sxml tree fold seems to not support sxpath.

<snape>I can't get append! to work on empty lists. Is that normal?

<ijp>yes

<ijp>most of the confusion around foo! functions on lists stems from the fact that lists aren't single objects in scheme

<snape>I see. But set! works on empty lists. That's where I'm confused.

<snape>I'd expect the append! implementation to kind of re-use set! if its first argument is an empty list

<amz3`>snape: you rather not use ! procedures for the time being

<amz3`>fwiw I never used append!, but I do use set!

<ijp>snape: I think you are also a little confused about set! fwiw

<amz3`>also mutation doesn't necessarly provide optimisation afaik

<ijp>try writing out your idea using set! and see if you realise why it can't do that

<amz3`>ijp: fwiw, I created a small editor using your fingertrees datastructure :)

<ijp>ah cool

<amz3`>it's less featured than guile-on-zile tho

<snape>ijp: my idea works with set!. I wanted to do (append! list list2) instead of (set! list (append list list2))

<snape>to avoid the 'list' repetition

<ijp>snape: I mean write append! to use set!

<snape>ijp: that would be a bit overkill :)

<ijp>anyway, you are supposed to use set! with foo! on lists

<ijp>you can't get around it, in general, because lists are not a single object

<snape>yeah, got it

<snape>thanks

<rekado_>I’m now using pre-post-order, which is a bit heavy for my use case, but works fine.

<amz3`>rekado_: what are you talking about?

<AndroUser2>Hello

<amz3`>welcome AndroUser2

***AndroUser2 is now known as YellowSphere

<artyom-poptsov>YellowSphere: Hi!

<sneek>Welcome back artyom-poptsov, you have 1 message.

<sneek>artyom-poptsov, civodul says: here's the fix i used for Guile-SSH in Guix: https://git.savannah.gnu.org/cgit/guix.git/tree/gnu/packages/ssh.scm#n230

<artyom-poptsov>sneek: Later tell civodul: Thanks, I fixed that bug -- see my email.

<sneek>Will do.

<civodul`>thanks artyom-poptsov

<sneek>civodul`, you have 1 message.

<sneek>civodul`, artyom-poptsov says: Thanks, I fixed that bug -- see my email.

<YellowSphere>artyom-poptsov: I'm the one who does shy project

<artyom-poptsov>YellowSphere: I see. Welcome to #guile!

<artyom-poptsov>Feel free to ask Guile/Scheme related questions here.

<amz3`>YellowSphere: what is shy project?

<artyom-poptsov>amz3`: We (I and YellowSphere) are trying to write a Bash lint (a static analyzer) in Scheme: https://github.com/artyom-poptsov/shy

<artyom-poptsov>It's a work in progress though, no stable version yet.

<YellowSphere>amz3`: my project's program checks deprecated syntax of .sh files

<amz3`>oh!

<amz3`>I am a bash noob

<YellowSphere>Just small program that checks syntax.

<amz3`>a linter of some sort?

<YellowSphere>Something like that

<spk121>Good morning, Guile. Here's a question. I needed a procedure to find the name of the current procedure. Here's what I came up with. http://paste.lisp.org/display/347217 Is there a better way?

<amz3`>we need something like that for Guile too

<amz3`>maybe you will come up with a framework for building linter in scheme :)

<ijp>I don't think __FUNC__ needs to be a macro, although the other two might need to be

<ijp>of course, if you do it as a function it might give weird results if it were ever tail-called

<artyom-poptsov>civodul: There are two segfaults during Guile-SSH tests with Guile 2.2 (I perform testing in a Docker container <https://hub.docker.com/r/avvp/debian-guile/>). Currently trying to pinpoint the cause of the errors.

<ijp>spk121: the only real improvement might be to make them identifier macros

<spk121>ijp: Thanks. "identifier macros"? I guess I'm unfamiliar with the term.

<ijp>spk121: (info "(guile) Identifier Macros")

<ijp>essentially it would let you use __FILE__ rather than (__FILE__)

<ijp>it might be possible to do __FUNC__ statically using syntax parameters, but my macro-fu is rusty

<ijp>but I think that would involve redefining define, which is not what we want

<spk121>ijp: Cool. Thanks again.

<ijp>I suppose the consolation is that the loop is probably never going to be more than a few iterations in practice

<amz3`>mwette you here?...

<amz3`>brendyn: what's up?

<janneke>i'm wondering about non-pre-compiled info

<janneke>would it be possible to have instead of (not compiled):

<amz3`>davexunit: did you have a look at my frontend framework https://amirouche.github.io/forward.scm/ ?

<janneke>instead have something closer to (compiled):

<brendyn>amz3`: Not much, just been doing a little bit more of SICP

<amz3`>ACTION did not read SICP yet.

<amz3`>I must be the only one here

<spk121>amz3`: nah. I never got around to it. But I was never much of a student of compsci. Just a poor engineer trying to make the computer do the things...

<amz3`>spk121: me too :)

<brendyn>Well I have to learn from somewhere since i self teach

<amz3`>brendyn: you don't prior CS knwoledge?

<brendyn>Nope

<amz3`>brendyn: then it makes sens you read SICP

<brendyn>There are many things even in the first few chapters/lectures that are profound and interesting, and they show you how to write good software instead of just mashing stuff together with guesswork

<amz3`>catonano: o/

<catonano_>amz3`: ehy ;-)

<amz3`>catonano_: doing good?

<catonano_>well...

<catonano_>I realized that packaging stuff for guix is daunting ;-)

<amz3`>daunting?

<catonano_>it' s ok, really. Just jioking ;-)

<catonano_>yes, daunting

<amz3`>I don't daunting, /me looking up wiktionary

<catonano_>lots of litte things to know

<amz3`>catonano_: yeah! completly correct statment

<catonano_>a lot of work