<dsmith>Hey! It built. With only a few changes. Two functions difference. A needed #include, and updating configure for 3.0 <mwette>I'm guessing scwm is a window manager in scheme. There used to be sawfish (aka sawmill). <ZombieChicken>I'd assume so, too, but there are so many acronyms, I'd rather ask and be sure <daviid>str1ngs: no problem, let's see if it is a 'real' fix, i think so but ... <str1ngs>daviid: looks good so far. thank you. <daviid>str1ngs: welcome, tx for the report <str1ngs>daviid: it was an odd bug for sure. webkit has strange IPC though. was that related? <daviid>str1ngs: no, it was a very big bug of mine! <daviid>str1ngs: such a bug that i deserv a brown bag for at least a week :):) <str1ngs>daviid: bugs happen :). though in all fairness events have worked quite well since inception. I thought I might find more bugs by now. <dsmith>ZombieChicken: Yes, Scheme Configurable Window Manager. <daviid>str1ngs: wrt to the example 'per se', you know you could use eq? (rather then equal?), then you may even use case, which would, imo, make the code 'cleaner', like this - https://paste.debian.net/1126732/ <daviid>str1ngs: the case for will also allow to treat more then one load-event at the same time - ... (case load-event ((started redirected) <your code here ...>) ((commited) ...)) ... <daviid>str1ngs: just some thought ... fwiw <rlb>wingo: in v3.0.0 it's still libguile/libguile-2.2-gdb.scm? (Fine with me, just noticed it while working on the debs.) <mwette>rlb, wingo: when I install 3.0.0 the only "gdb" file I see installed is libguile-3.0.so.1.0.1-gdb.scm <apteryx>erkin, dsmith, erkin: thanks for the inputs regarding my question about "partitioning" :-). I'll have a closer look at what's in srfi-1. ***Server sets mode: +nt
<civodul>do people see segfaults while building Guile 3 on ARMv7? *civodul does but it might be non-deterministic <jonsger>civodul: I see a different then you. Unrelated <civodul>jonsger: but you do see a segfault while it's building, right? *civodul tries to get a backtrace <civodul>so yeah, SIGILL might be because you're not on the right arch or something <dsmith>civodul: I did not get a segfault on armv7l gcc 6.3.0 ***nerdypepper is now known as nerd
***nerd is now known as np
***np is now known as nerdypepper
<civodul>would have been too easy it everyone had the same problem ;-) ***nerdypepper is now known as n
***n is now known as nerdypepper
<dsmith>civodul: I would suspect a different compiler version handling something "unspecified" in a different way. <sneek>Sneeky bot running on Guile version 3.0.0 using bobot++ 2.3.0-darcs <dsmith>civodul: The bot is on 3.0 on arm <jonsger>wingo: do you have a guile 3.0 compatible version of guile-charting? ***ng0_ is now known as ng0
<lloda>guile-cairo does work on Guile 3 w/o issue <lloda> other than maybe the install, I haven't checked that <str1ngs>funny I was just going to see if I could map out some goop classes with g-golf and cairo <daviid>str1ngs: cairo is not introspectable, but we have guile-cairo ***apteryx_ is now known as apteryx
<str1ngs>darn I was hoping to use it to draw things with g-golf <daviid>str1ngs: it is possible, maybe it needs some special treatment, but that has been done for guile-gnome and gle-clutter - so they can use a cairo surface ... <str1ngs>gobject-introspection ships with cairo-1.0.gir but I think that is just providing some type information. <str1ngs>maybe clutter would work just as well <daviid>str1ngs: no, cairo is a complete and necessary 'tol' for it offers - clutter is a complete diff best <daviid>*complete different and necessary tool <str1ngs>I've used clutter before, cairo would be better for what I need. I just want to map out goops class hierarchy <daviid>str1ngs: you generally need both, in my experience <str1ngs>I think I can draw to a GtkDrawingArea <str1ngs>but without the cairo GI. i'd have to do the drawing in C <daviid>str1ngs: i don't have time now, but everything can be done in scheme, i just posted an example and the code ... in gtk4, clutter is included in gtk ... <daviid>have to hack o somethng else now... bbl <daviid>this none truncat excaption is a disaster <rekado_>trying to build pfds with Guile 3 but get “Unbound variable: &assertion” in procedure “raise-exception”. <rlb>With 3.0.0 00-repl-server test is failing here with fport_read EINVAL (in I think the read-line) when I build and run it as root, but not when run as another user. Any idea why that might be? <rlb>(and of course I don't know that "root" is the critical difference, just that's when it appeared, and it's completely reproducible -- happens every time) <rlb>...looks like it's the ptob->c_read in scm_i_read_bytes that's throwing the EINVAL. <rlb>...odd -- at that call ptob->c_read is neither a pointer to read() nor to trampoline_to_scm_read, and I can't see any other places c_read would be set to any other value. <rlb>(though I'm guessing that's not the issue...) <rlb>Current guess is either something's wrong with the repl socket when building/running as that other (root) account, or there's a race that's plays out differently there... <rlb>Oh, well -- I'll come back to this, but looks like 3.0 debs will be delayed a bit longer. <rlb>civodul: ahh -- I noticed you'd had to deal with some 00-repl-server test issues in the past, and right now I'm finding that the test reliably fails when I build 3.0.0 from the v3.0.0 tag as root, but not as a normal user. (Could well not be "root" related.) It fails with an fport_read EINVAL in the read-line in read-until-prompt in the "simple expression" test every time. <civodul>rlb: that doesn't ring a bell, but i wouldn't run tests as root :-) <rlb>They'll always run as root when the debs are building on the buildds. (Debian packages have to be built as root -- or fakeroot.). <rlb>But no, I don't normally do that, day to day... <rlb>I traced it down into scm_i_read_bytes -- the ptob->c_read() call there, but no idea why... <rlb>also, not sure this is relevant, but the c_read pointer there isn't either read or the trampoline_read -- it's some other address. <rlb>(couldn't see how that could happen) <rlb>(I mean at that point -- I printed them via %p) <rlb>But I'm sure c_read could be getting assigned some way other than the obvious ones... <rlb>Anyway, I'll have to figure this out before I can finish the 3.0.0 debs. <akhetopnu>has anyone here used websockets in guile? what library did you use? is it available through guix? <outtabwz>looks like there is interest in guile emacs <rlb>I'm certainly interested, but as mentioned there, I suspect who really has to be(come) interested is the Emacs maintainers. <janneke>it's packaged in guix, but yes it's important to get emacs hackers "on board" <outtabwz>some people are very protective of elisp <outtabwz>elisp and scheme are both considered lisp1 right? <jcowan>That is, an identifier can be bound to both a value and a function, and (foo x) uses the function binding. If you want to use the value binding, you have to write (funcall foo x). <jcowan>There are some Lisp-1s that aren't Scheme, but I can't think of any offhand. <outtabwz>i think there are some other things though <RhodiumToad>including things like magit, which is extremely useful <outtabwz>it would be neat if guile's elisp module could handle all that' <zig>in particular, IIRC the last two commits are garbage. <dsmith>Wooo. Got finally got the scwm simple.scmrc working. <dsmith> define*-public doesn't curry any more. <rekado_>It seems that (rnrs condition) has a bug: it exports &assertion but only imports &assertion-failure as &assertion-violation, so &assertion is actually undefined. *rekado_ submitted a bug report <mwette>Can someone define what Guile Emacs is? Is this guile used as emacs-engine running elisp, but also providing bindings in scheme (for the stuff in emacs lisp books)?