IRC channel logs

<rlb>sneek: later tell civodul any idea offhand why the recent signal thread changes might cause the standalone tests to hang on test-guild-compile with the parallel test harness? May be repeatable for -j4 and greater here...

<sneek>Will do.

<rlb>(reverting them "fixes" it...)

<civodul>rlb: hi! re sigaction test, i have no idea why it would make tests hang

<civodul>can you see which process “make” is waiting for?

<lloda>civodul: wdyt about making SCM_DEBUG_TYPING_STRICTNESS default to 2. There's a comment saying this is inefficient, but i find that hard to believe

<civodul>lloda: yeah, i doubt it can be inefficient (maybe we can build with and without and compare the stripped binaries to see if it changes anything at all)

<civodul>making it the default when building Guile itself is prolly a good idea

<civodul>(i don’t think we can change it for Guile users, at least not in 3.0)

<sneek>chrislck: wb!

<lloda>ok. However i think this would have no impact in users, except in revealing bugs perhaps

<lloda>on users i mean

<lilyp>"Revealing bugs" is a big impact for those who consider them features :)

<lloda>i was forced to use STRICTNESS 2 for my own code bc c++20 rejects the volatile trick in STRICTNESS 1

<lloda>the bugs i found were all using SCM as int or pointer and viceversa

<lloda>mostly bc of using one of the scm_xxx_p as if they returned a C bool

<lloda>which is very dangerous

<civodul>it is

<sneek>wb dsmith!

<dsmith>sneek, botsnack

<sneek>:)

<dsmith>!uptime

<sneek>I've been serving for 2 months

<sneek>This system has been up 42 weeks, 4 days, 21 hours, 7 minutes

<old>Almost a year!

<old>That is some robust bot

<chrislck>sneek: botsnack

<sneek>:)

<chrislck>\o/

<graywolf>Hi :) I am trying to wrap my head around call-with-prompt, and I have troubles understanding the handler argument. I understand the wording in the docs, but only example has (lambda (k) k), which is not very helpful.

<graywolf>Would people have examples for what/how the handler is supposed to/can be used?

<lilyp>try (lambda (k) (sleep 5) (k)) – this should resume your computation after 5 seconds

<dthompson>graywolf: one simple example is stashing k in some variable for use later.

<dthompson>or yeah what lilyp just said

<lilyp>note that k can be picky about the number of args it takes

<dthompson>graywolf: example of storing k somewhere (replace 'yield' with 'abort-to-prompt' in your head) https://git.dthompson.us/chickadee/tree/chickadee/scripting.scm#n52

<graywolf>Oh, I see, thank you both :)

<cow_2001>i need to learn how to use the debugger. i'm being crushed by all the prints

<cow_2001>help help i'm being oppressed!

<cow_2001>in a better world people who make bad python references get electrocuted

<rlb>civodul: ok, thanks, and haven't had a chance to delve yet -- initial guess was something load-related, so I thought I'd see if that rang any bells. i.e. at -j4+ here, all the (4) cores are more likely to be busy, etc.

<old>is there a planned release soon for 3.10?

<civodul>we should do that, but i don’t think there’s a plan

<old>okay

<old>I'm saying this because I have a patch waiting since October 2022 and hope that it will make it in the next version ^^

<civodul>ouch

<civodul>it’s in the bug tracker?

<old>not sure

<old>I posted it on bug-guile@gnu.org

<civodul>great

<civodul>this: https://issues.guix.gnu.org/61058 ?

<old>hmm I had this in mind: https://lists.gnu.org/archive/html/bug-guile/2022-10/msg00020.html

<old>but they seem the same I think

<old>Just a different approach

<old>but yeah typically a deadlock on join-thread

<civodul>both are by you, right? :-)

<old>yes, I think I got confused wheter I need to open a bug or post a patch on the ML

<civodul>i’m happy to apply https://issues.guix.gnu.org/61058

<old>or maybe I though that the nobody was reading the mailing list so I made a bug report in 2023 :-p

<civodul>heh

<civodul>a test case would be nice

<civodul>but maybe tricky?

<old>entirely possible I can do it

<old>but I think it would need to be protected by dynamic-wind

<old>wait-condition-variable will execute async marks when timeout IIRC

<old>it is possible to throw in the async procedure, thus by-passing the unlock of the mutex

<old>hmm not true. only on EINTR

<old>so that's fine

<civodul>yeah, nothing can throw in ‘lp’ as i see it

<old>how would one test this however? I guess the current behavior that is observed should not happen anymore

<civodul>right

<civodul>you write a test that reproduces the bug

<civodul>i guess it’s enough to do (join-thread t 0) followed by (join-thread t)

<civodul>right now the second call just hangs forever, right?

<old>hmm no that's fine

<old>(let ((thread (begin-thread (pause)))) (join-thread thread 1) (join-thread thread)) => 'misc-error "lock-mutex"

<old>so that's very simple in fact

<old>but you do want two timeout to avoid the hang for ever

<old>civodul: I'm going to make a v2 with the test in it

<civodul>👍

<old>hmm how does one handle the case of Fedora with "guile3.0" in scripts?

<old>what a pain is this distro ..

<civodul>ah, well

<old>civodul: sent!

<old>eeh I just saw that there's a mixture of space and tab for indentation in the file. The patch does not look clean because of that

<civodul>ACTION looks

<civodul>old: pushed!

<civodul>we made progress tonight

<ieugen>hi, are there any up to date GUI libs for guile? I would like to try my hand in making a NetworkManager plugin

<ieugen>I saw https://github.com/spk121/guile-gi - does not seem that up to date, and I might require gtk4

<ieugen>hm I will probably need a dbus library as well

<old>civodul: thank you!!

<old>civodul: now I was wondering, when's fiber going to be core guile? :-p

IRC channel logs

2024-01-25.log