IRC channel logs

<lechner>lampilelo: thanks!

***Server sets mode: +ntz

<chrislck>sneek: botsnack

***Server sets mode: +ntz

***roptat is now known as Guest8848

<ulfvonbelow>(file-port? (car (pipe))) => #t

<ulfvonbelow>is `file-port?' supposed to indicate whether a port is related to a file, or whether it's related to a file descriptor?

<dcottingham>Pretty sure file-port? is about whether it's a file. Also pretty sure all ports have associated file descriptors. But maybe it's time for an experiment.

<dcottingham>Nope. New theory: not all ports have associated file descriptors, and file-port? is about whether it has a file descriptor.

<fcw>Question: Why is SIGINT and SIGQUIT ignored in (system* "bash" "-c" "trap") but not in (system "bash -c trap") ?

<jpoiret>fcw: maybe because system uses bash under the hood to launch the command, while system* does not?

<fcw>jpoiret: I suspect that it is caused by these lines https://git.savannah.gnu.org/cgit/guile.git/tree/libguile/posix.c?h=v3.0.8#n1524 but I don't understand the reason for ignoring these two signals.

<jpoiret>apparently, that's because you may want Ctrl+C on the foreground process to only reach the child, so that the parent can handle the failure of the command

<fcw>Oh well, looks like I'll have to use `system` instead of `system*` then.

<fcw>jpoiret: By the way, I encountered this "problem" in Guix. Mailing list thread: https://lists.gnu.org/archive/html/guix-devel/2022-05/msg00209.html

<jpoiret>right, i saw that. I wonder how the executable could inherit those though

<fcw>I was wondering the same, but I have not found the cause.

<sneek>wb chrislck :)

<dsmith-work>Tuesday Greetings, Guilers

<nckx>jpoiret: Processes always inherit signal masks by default.

<jpoiret>yeah but from what i understand, it's the resulting binary that has the same behivour

<nckx>I thought you were talking about bash.

<nckx>I mean

<nckx>this ML stuff is just

<nckx>insane.

<nckx>Is it a run/dump image/restore kind of language/compiler? That would be my guess. That it just serialises its entire $state.

<nckx>Utterly bonk.

<jpoiret>ah right, didn't think of that

<jpoiret>tex style

<lechner>Hi, are there libpam binding (as in Linux PAM) for Guile?

<lechner>bindings

<lampilelo>lechner: here are two lists of guile-based projects:

<lampilelo> https://www.gnu.org/software/guile/libraries/

<lampilelo> http://sph.mn/foreign/guile-software.html

<lampilelo>i doubt there are pam bindings for guile anywhere though, you can always use guile's ffi

<lechner>lampilelo: thanks!

<lampilelo>lechner: also you can (more or less) automatically generate guile bindigs for any c library using nyacc (https://www.nongnu.org/nyacc/) or or swig (https://www.swig.org/Doc4.0/Guile.html)

<lechner>lampilelo: thanks for those, too!

<lampilelo>nyacc generates ffi bindings written purely in scheme whereas swig will create c code to be compiled to a dynamic library you can then load as an extension

<lechner>lampilelo: yeah, i'll have a look at nyacc first

<lechner>Hi, why do I get guild: unknown script "compile-ffi" when trying ti use nyacc on Guix, please?

<lampilelo>hm, i don't remember how scripts are loaded, maybe you need to have it in the load-path? or maybe you can pass its absolute path to guild

<dsmith-work>sneek: botsnack

<sneek>:)

<mwette>if the nyacc module/ dir is in your GUILE_LOAD_PATH then "guild compile-ffi -h" should work

<sneek>mwette, you have 1 message!

<sneek>mwette, civodul says: a stop-the-world GC is indeed triggered when libguile makes an open(2) call that returns EMFILE

<mwette>"guild compile-ffi ..." will result in loading the module (scripts compile-ffi)

<lechner>mwette: thanks! that's my issue. the module is not available in /run/current-system/profile/share/guile/site/3.0/

<stis>hello guilers!

***Server sets mode: +ntz

IRC channel logs

2022-05-24.log