IRC channel logs
2022-02-08.log
back to list of logs
<civodul>daviid: hi! i think that test is a bit flaky, but that's not a regression <daviid>i am gona install and run a mini benchmark on g-golf, if g-golf still works :) - i am curious if the string->symbol patch makes things better for g-golf as well <civodul>it has a slight effect on the compiler because it fiddles a lot with symbols <civodul>but for "regular" code it's probably hardly noticeable <daviid>civodul: ok, g-golf spends its life calling it at import (typelib) time, but not to run user code... importing Gtk 4.0 takes 6s on my notso slow laptop ... <daviid>of course users should selectively (only) import the classes they need, but I was interested 'benchmark' a full import and compare 3.0.7 with 3.0.8,let's see <stis>playing with stacks and pausing continuations. e.g. <stis>I want to benchmark if we can get python on guiles generators to move any faster by not capture the stack, but run it on a separate stack <vijaymarupudi>Is there going to be a recording of the Spritely Goblins talk in FOSDEM? I was looking forward to that talk <rlb>Certainly, and you can see some other tests we've muted for similar reasons. <rlb>(in case they're relevant) <rlb>Though some may be sensitive to the architecture. <daviid>rlb: ok tx - i have been interrupted half way through, but i shall now install and try g-golf against main ... <dsmith>rlb: There was some test that was failing for me when I upgraded to bullseye, but it seems to be ok now. (Can't remember what it was). <dsmith>rlb: Do you ever go back an re-enable disabled tests to see if they are still needed? <rlb>dsmith: not systematically, but ideally, I *should* do it for each release. *rlb makes a note for 3.0.8, though it might make more sense for X and Y releases unless we expect things to have changed for some reason. <mwette>hey rlb, btw, really appreciate you keeping guile and debian well mated <rlb>thanks - happy to, though I need to do some catch up soon. <daviid>just had a series of X11 crashes i never ever had in ... decades, i just updated my debian testing ... and now, starting emacs, then hit C-x-52, at a certain moment, that action crashes X - and all applic die, terrible <daviid>tll the next crash, ,,, i am here again <daviid>it crashed again - ma machine became unusable <rlb>daviid: not sure if you saw it, but I said "Hmm, might check https://bugs.debian.org/emacs and see if anything looks familiar. I think I likely need to catch up there too." <daviid>rlb: i was trying tolook, both emacs and epiphany crash every second or so, themachine just isunusable <rlb>Hmm, sorry. That's not good. <daviid>very frustrating, but... once every 20y or so... <rlb>I'm not on current testing, though I have one machine that's "close" (within weeks). <rlb>But I'm mostly sway+firefox so likely very different envt. <daviid>debian stable is unusable 'in real life', only for those who can offerto live 5y ago permanently <rlb>fwiw, on my main machine I generally use a pinned stable setup, i.e. mostly stable, but with the ability to pull in bits from testing/unstable when I need to, and over time, that drifts more toward testing (between stable releases)... <daviid>rlb: which bug should i look in particular, <rlb>Oh, I don't know - I'd probably just scan the most recent ones... <rlb>(i.e. I didn't have anything particular in mind) <rlb>Hmm, interesting. I use emacs-lucid to avoid the long-standing gtk bug, but this sounds like something different... <rlb>i.e. I still use emacs in --daemon mode on remote servers for some things, though I don't generally use X forwarding anymore. *rlb wonders if we should move this elsewhere -- if so, feel free to /msg me directly. <daviid>yeah, let's not 'invade our channel',tx <daviid>the good news is 3.0.8 is substantially faster then 3.0.7, wrt g-golf import, nearly 2x faster <mwette>emacs crashes for me all the time; but I'm on ubuntu running wayland <rlb>mwette: gtk or lucid? <daviid>mwette: debian runnig wayland, i never had an emacs crashes in years, now ... every second or so :):) <dsmith>daviid: Oh that is so frustrating. <dsmith>daviid: When Debian moved to systemd, my laptop would only run for about 30 seconds or so, and then power off. It was kind of like "Groundhog Day" trying to figure out what was wrong. <daviid>dsmith: yeah, i am scared to hit 'new frame' now, which seems to cause the crash, not 'always' though - let's see - i wana paste my 2 'mini benchmark' <dsmith>The machine was reporting the lid being closed, so systemd was doing the shutdown stuff. Fixed by diabling checing for the lid close. <rlb>wrt that bug, do you have debug-on-error set? <daviid>rlb: no, it just started to happen 'now', so i am trying to figure a way to 'not crash' and still work ... a bit :) <daviid>mwette: fwiw, following an advice from rlb, i am now using emacs-lucid, which while installing purged emacs-gtk, and everything seems smooth so far - <rlb>i.e. why I still use emacs-lucid <rlb>though I haven't checked the upstream status there recently <daviid>rlb: unfortunately, emacs-lucid just did crash twice in a row <daviid>itis literally impossible for me to work now <daviid>all of a sudden, emacs crashes and crashes several app with it ... <rlb>daviid: suppose you might try installing xfce or kde and logging in with one of those sessions to try to check whether it's gnome-specific. <rlb>...though if you don't want those packages installed, it might be a bit tedious to remove them later. <rlb>I sometimes keep them around for just such purposes (particularly xfce) <daviid>isodepend on 3 things to work, gnome, emacs and epiphany <daviid>the rest can 'die', butthose 3 coponents, then i die :) <daviid>i thinki have xfce4 i'll try now <dsmith>How could emacs crashing cause other apps to crash? Wayland maybe? <gnoo>dsmith: try with sway and a minimal config <gnoo>login to a virtual tty and type sway, no need to install it systemwide or fiddle with gdm <dsmith>gnoo: Oh, I'm fine. daviid is having the problems. <daviid>right, tx for the help and tips, sorry for so manyot msgs <dsmith>X11, awesome, emacs, xterm, firefox. No problems for me. <daviid>i temp starting a gnome classic desktop, which backen uses X11, and so far 'ok', soi think it is a wayland problem, not an emacs problem ... we'll see *dsmith looks squinty-eyed at weyland <ArneBab>daviid: I switched to X11 with exwm a few weeks ago, because Gnome clobbering my keyboard layout when changing users and Chromium messing up keybindings when switching back got too annoying. <daviid>wingo, civodul: here is what i meant to paste yesterday :), if i didn't have this terrible xwayland bug (i am now running gnome on xorg) - which are 2 g-golf profiles, using 3.0.7 and 3.0.8, showing that 3.0.8 is nearly twice as fast as 3.0.7 on the job - https://paste.gnome.org/pyrq9eaux <dsmith>Abount the same amount of time in GC <daviid>i must be cursed or something, now in gnome xorg, all of a sudden my mouse left and middle click do not work anymore <daviid>which makes the desktop env unusable again ... <cwebber>vijaymarupudi: yes there will be a recording <cwebber>I can send you a copy of the recording without Q&A if you'd like <mwette>daviid: thanks, I will look into that <cwebber>however these don't include the (very interesting) Q&A sessions that happened <m455>oh is there a list of ice-9 modules somewhere? <m455>i was meaning to ask this a few days ago, and then forrgot haha <m455>or do you just sift through the manual to find them? <m455>ahhh thank you both of you! <3 <m455>ahh this is so much better than the extremely vast paragraphs i go thoruhg in the manual hahah <m455>having a list of all of the libraries and their functions in a terse manner would be a great improvement :O <m455>this will do though!!! <3 <m455>i understand that people do all of that documentation on their free time though, so no hard feelings on them either! <dsmith-work>Has anyone ever been paid to work on guile? (Tom Lord, probably?) <lagash>dsmith-work: as in, hack on Guile itself, or just Guile programming? <lampilelo>kinda incredible, gnu projects rarely get financed but there's so many of them regardless <avalenn>is there any way to do pretty-print but for records ? <civodul>avalenn: not out of the box, because records don't have "read syntax" (that is, you cannot read them with the 'read' procedure) <dsmith-work>Is there a way to use goops for that? Would format be smart smart enough to call a write|display|pretty if one was implemented for that type? <avalenn>Ok. I will look at what I can do for my use. <lilyp>You can set a record printer, but you'd need to read them back somehow