IRC channel logs

2021-10-12.log

back to list of logs

<daviid>lloda: how do you 'usually' apply your exception-format patch, or maybe how would you recommend users to do so? do you have a script that, i.e., checks thesuer version, grab the 'good' ice-9/boot-9 using wget or curl, apply locally, recomnpile instyall that file only ...?
<daviid>lloda: i am comnparing the ice-9/boot-9 from 3.0.7 to the one in wip-exception-truncate and get 'scary' results, fwiw
<daviid>hum, i must jhave done something wrong to grab the proper ice-9/boot-9 from our branch- will retry :)
<daviid>lloda: i think your wip-exception-truncate branch is 'out of date' with main, if you could confirm and possiblyu update your branch, many thanks
<cwebber>hello
<drakonis>hey there.
***jpoiret5 is now known as jpoiret
***herlocksholmes9 is now known as herlocksholmes
<cwebber>just sent an email to guile-devel
<cwebber>I'm still hoping we can get ijp's javascript generating branch merged.
<dsmith>sneek: botsnack
<sneek>:)
<dsmith>sneek: botsnack
<sneek>:)
<dsmith>!uptime
<sneek>I've been running for 46 seconds
<sneek>This system has been up 11 weeks, 5 days, 10 hours, 44 minutes
<dsmith>!uptime
<sneek>I've been running for 6 minutes and 53 seconds
<sneek>This system has been up 11 weeks, 5 days, 10 hours, 50 minutes
<lloda>daviid: it's outdated, i'll rebase
<daviid>lloda: great, thanks
<lloda>i think there *is* a proper solution to this without the patch, just have to spend a bit thinking on it
<daviid>lloda: oh, that would be fantastic- i am working on a guile-cv release exactly now, and just updated the manual with these 'manual' steps users must perform to be able to use guile-cv... very depressing imust say, and a proper solution would be very wekcome! ping me whenever you had time to ... and tx!
<lloda>sure. the patch is nasty :-\
<lloda>daviid: just pushed rebased wip-exception-truncate
<daviid>ok, tx
<stis>o/
<stis>o/
<dsmith>!uptime
<sneek>I've been running for 10 hours
<sneek>This system has been up 11 weeks, 5 days, 21 hours, 18 minutes
<ArneBab>wingo: is there an important reason why for fibers #:parallel is by default #f? What’s the usecase in which that is better? (I spent a few hours debugging this weekend before realizing that I forgot to add #:parallel #t).
<ArneBab>wingo: aside from that: Thanks to fibers I now have a server-driven websocket interface that allows me to provide a commandline storygames in the browser.
<ArneBab>still need to implement session-functionality (currently it is a single-user browser-game, which would be a bit unexpected: “sorry, you've got to wait, someone else is playing on this website right now” :-) )
<ArneBab>#:parallel?
<stis>ArneBab: Do you know how to interact between fibers and g-golf?
<stis>I use now different threads and mutexes that fro mthe fibers side is guerded with a non asynch chunk
<stis>wingo: any aideas ^^
<ArneBab>stis: I don’t, no … for my game I had to skip the soft-ports. They would have been the most elegant API, but they are a continuation barrier.
<ArneBab>all in all the fibers are pretty cool, because they provide multi-threading without its complications.
<ArneBab>but I use far too little of their functionality: Guile can handle a million fibers and I use just 5 :-)
<ArneBab>But for async interaction the "put and get messages via channels"-model is very nice.
<stis>thanks!
<ArneBab>it would be pretty cool to have a GUI library that relies completely on fibers for synchronization: many independent components that are only connected via channels.
<ArneBab>wingo: is it intentional, that yield from (ice-9 threads) provides a yield-point for fibers? (it’s not in the manual, that’s why I ask)
*stis having fun with g-golf
<dsmith-work>Tuesday Greetings, Guilers
<dsmith-work>I wired up the bot last night with systemd, so it *should* restart if it crashes.
<chrislck>it's now an autobot
<lampilelo>is it a russian bot?
<lilyp>sneek tell lampilelo нет
<sneek>lampilelo, lilyp says: нет
***chris is now known as chrislck
<cwebber>hello #guile!
<cwebber>rekado_: wingo: hi there... so in addition to looking at ijp's js branch, I'm looking at robin's stuff, looking from where rekado_ last rebased
<cwebber>there seem to be two kinds of commits that are in here: some were doing some lower-level compiler optimization tweaks for speed, some were stuff specifically for elisp
<cwebber>right now, the merge for some of the lower-level stuff is.... difficult
<cwebber>it's hard for me to know how to reconcile those tweaks with stuff that has happened since
<cwebber>so I might just see if I can cherry-pick the elisp specific stuff
<cwebber>and leave the rest of the commits as something to explore afterwards
<cwebber>wdyt rekado_, wingo ?
<robin>cwebber, did someone merge the branches? i'm slightly confused about the context
<cwebber>oh hai robin
<cwebber>:)
<robin>(and many thanks to you and, iirc, rekado_ for cleaning up my messy repo)
<robin>hi cwebber :)
<cwebber>robin: I had done a rebase a few years back, and last year rekado_ did another one https://git.elephly.net/?p=software/guile.git;a=shortlog;h=refs/heads/wip-elisp
<cwebber>robin: https://git.elephly.net/?p=software/guile.git;a=commit;h=b04e79283ada9a6af05552dda6446a0934c0fbe2
<cwebber>this one I just don't know how to resolve with master
<cwebber>and there's a few others like that
<cwebber>however all the ones that are touching stuff in modules/language/elisp look, at first glance, relatively easy to merge
<cwebber>I wonder how dependent they are on the earlier commits
<cwebber>or if those are mostly performance tweaks (which may now be replaced by very different performance tweaks by wingo)
<cwebber>I'm not sure, I'm thinking out loud
<cwebber>robin: any insights you have are most welcome tho :)
<cwebber>robin: my gut sense here: you and ijp did two awesome projects to bring wonderful things to guile's compiler tower
<cwebber>and without them being merged, they keep birotting and becoming incompatible with master
<cwebber>I'd like to fix that, get them merged asap so they can be maintained along with the rest
<robin>yeah, it was quite silly of me to leave the last bits unmerged. i'll have to go back and remember/reconstruct why i did certain things...
<robin>arbitrary constants, hm. i don't recall the context, but it seems to only care about them when the assembly's not going to be dumped to a file; i'd guess some elisp does the equivalent of (eval `(lambda (...) ',value)) where value is unserializable (which might be...what, arbitrary structures maybe?)
<robin>maybe i can dig up my old notes (which might be useless, who knows)
<robin>i'll probably have some free time in the next few weeks to look over it more closely. and of course i know very little about what the guts of the compiler look like these days
<cwebber>robin: that would be awesome if you could review / see how to get on top of master
<cwebber>or, main
<robin>i guess i could also take out the questionable code and see what breaks...
<robin>this doesn't look too bad overall except for the constant-interning stuff
<robin>and yeah, ijp's work is very impressive too :) and useful, since WASM seems to be moving forward a *bit* more slowly than hoped with things like gc integration (possibly it just landed experimentally last year?)
<cwebber>robin: yeah
<cwebber>robin: and good re: doesn't look so bad! yes I was thinking the stuff touching module/language/elisp looked quite easy to merge
<cwebber>I just don't know what to do with that interning stuff
<robin>cwebber, did compiler internals change enough to break it significantly?
<robin>(i mean, from your comments that seems obviously the case, but...)
<robin>i'll probably revert them and wait for a guile-emacs crash, my best guess is elisp programs written to be run by an interpreter (and therefore being free to stuff completely random objects into code)
<cwebber>robin: I don't think it was large changes
<cwebber>I was just pretty confused with that one particular commit
<cwebber>cps changed significantly
<cwebber>not too much else did
<cwebber>but that one section did involve both you and wingo touching some similar areas I think
<dsmith-work>!uptime
<sneek>I've been running for 18 hours
<sneek>This system has been up 11 weeks, 6 days, 5 hours, 7 minutes
<ArneBab>chrislck: how does sneek look after transformation? :-)