IRC channel logs

2020-08-13.log

back to list of logs

<rilez>I wrote a rough draft of a tutorial on delimited continuations. Would anyone be interested in looking it over?
<rilez> https://github.com/rilerez/generator-guile-tutorial
<mwette>rilez: Thanks! I've saved the link for later enjoyment.
<rilez>woot feel free to file an issue or @ me if you have critiques!
<daviid>str1ngs, RhodiumToad: looking at your suggestion to 'make distclean' should not remove the configure file, I agree, will patch - while at this, could look at http://git.savannah.gnu.org/cgit/g-golf.git/tree/Makefile.am?h=devel#n180 and see if you spot other candidate for 'none removal', tx
<daviid>RhodiumToad: you mentioned 'at postgres we ...' are you a postgres core developer/maintainer?
<lisbeths>I have guix installed on debian and it is pulling right now so I know that guile is installed, but when I run guile it isnt in my path. It's somwhere in my guix path.
<rilez>hmm can you find guile in ur guixprofile
<lisbeths>hold on let me do another tty
<lisbeths>also I want to apologize for what I had said the other day on #guix I was off of my schizoid meds
<lisbeths>got me bant
<lisbeths>what directory is it in
<lisbeths>I found out I can actually can run this to get guile: guix repl
<nckx>lisbeths: Apology accepted 🙂
<lisbeths>The problem I have with the guix community, which I can now say with medication is as follows:
<lisbeths>I personally think guix is the future of package management, but my main computer with 16 gigs of ram does not have free drivers. I can not afford to go get a new laptop. Right now I have been compiling guix on a laptop with 2 gigs of ram so it will be supported. I think it is wrong to say that you can't help me sideload a nonlibre kernel onto my computer.
<nckx>OK.
<lisbeths>I have purged all of the non libre software from my computer painstakingly in order to appease the guix devs however the kernel is not one I can get rid of without extreme pain and time out of my own life.
<rilez>At least on my computer, .guix-profile is in my home directory
<nckx>lisbeths: You don't have to appease us (or anyone) at all. Your computer is yours to do with as you wish. Guix does not and will never contain a single line of ‘break proprietary drivers just because we can nye nye nye’ code.
<rilez>inside it i have ~/.guix-profile/bin, ~/.guix-profile/lib, and the like
<nckx>But that's very different from supporting (either in code or in IRC or elsewhere): that's *effort*. That's *time*. Nobody owes you that. There's nothing wrong with that.
<lisbeths>Well the question I had had was if I just swap out the kernel (which I can do myself) I don't know if the guix system will handle that properly.
<lisbeths>Debian handles things in such away that they have a more supported free version and a less supported nonfree repositories
<nckx>lisbeths: Sure. I do that.
<lisbeths>moreover the entire notion of "you can't even support the notion of proprietary software on your system" was brought about by stallman and I had thought that was what the whole debate surrounding all of this drama was about was ousting him.
<lisbeths>I am not trying to take sides merely to say that debian is a more expeirenced project and they separate nonfree and free out that way for a reason
<lisbeths>it is because not everyone can afford to go buy open source compliant hardware and they need drivers
<lisbeths>perhaps the person I was talking to previously was misinformed about what guix "supports"
<lisbeths>but I will take this discussion into the guix channel instead of here for less confusion
<rilez>Has anyone gotten guile running on webassembly?
<lisbeths>I think there are already some gpl 3 schemes that run on it. I am fairly certain. so that might be a good place to start, becuase alot of the same core functions will be compatible.
<lisbeths>or rather gpl 2 schemes
<rilez>cool, thanks! It probably makes more sense to shim the missing functionality than try to run the guille vm through emscripten
<rilez>I feel like the hardest thing to replace would be prompts
<lisbeths>you could also begin writing your app in a fairly r5rs compliant guile and "wait" for guile suppoort
<lisbeths>like 'expect' it
<lisbeths>similarly to how video game developers plan for their games to have twice as good graphics cards as when they start because it takes two to five years to make a game
<rilez>ahh yeah I'm still trying to work out what reasonably standard scheme should be
***terpri__ is now known as terpri
<rilez>Oh! Is there a way to make a callable object? Is my best bet to use a clojure? Is there a way to specialize a GOOPS method on it?
<str1ngs>daviid: hello I think anything created by autogen can stay, for example aclocal.m4 can stay same with m4/ . see distclean @ https://www.gnu.org/prep/standards/html_node/Standard-Targets.html#Standard-Targets particularly. ‘make distclean’ should leave only the files that were in the distribution. distribution as in tarball produced make dist-gzip .
<str1ngs>daviid: build-aux can stay too. to test you can do make distclean. then ./configure should still work.
***drakonis- is now known as drakonis
<str1ngs>rilez: I'm not sure what you mean by callable object.
<rilez>a struct that is also a function
<daviid>str1ngs: the reason it removes 'to much' is because I made 'make distclean' to work like 'git clean -dxf' (sort of ...) - users should use 'make clean' - but i agree that it can't remove to much, and configure must still work, that's a bug indeed, if it doesn't ...
<daviid>str1ngs: i wish i had a script to compare a fresh untar, then what is in after ./configure; make 'distclean' ... - so i could run it when i make changes to the 'dist' ..
<daviid>if anyone here has such a script, please ping me ...
<daviid>str1ngs: m4 is not entirely removed, afaict, but what is not in the tqrball
<daviid>anyway, i see it needs love ...
<str1ngs>daviid: I think instead of distclean you probably want maintainer-clean.
<daviid>str1ngs: imo, distclean should bring the dir back to its exact state as aftyer a fresh untar
<str1ngs>distclean is still useful for users since it ensure an configure created files are remove.
<str1ngs>daviid: that is right, that's my understanding of distclean as well
<daviid>str1ngs: yes, agreed - as it is, 'my distclean' removes to much
<str1ngs>no worries, maybe if you just use maintainer-clean that will avoid any issues with distclean.
<daviid>str1ngs: but i don't think the std distclean works as i wish either
<str1ngs>I hear ya
<daviid>which is why i enhanced it in the first place, a looong time ago :) - then got it too strong :):)
<str1ngs>rilez: does this help. http://paste.debian.net/1160238
<str1ngs>rilez: obliviously my name method is redundant. all I could think to use here.
<str1ngs>daviid: it grew too powerful! :P
<daviid>str1ngs: yes, i got the 'git clean -dxf' virus :)
<str1ngs>I like git -xfd too
<str1ngs>though for projects I don't maintain. I'll use make distclean
<str1ngs>just with git -xfd make sure you stage new files first. don't ask me how I know this :P
<str1ngs>err git clean -xfd
<daviid>str1ngs: yes, git clean -dxf can be dangerous ... we've all been there i guess
<str1ngs>thank goodness for emacs buffers. saved my bacon :)
<daviid>yep, emacs is a good os :)
<str1ngs>you know in emacs you can undo auto reverted buffers
<RhodiumToad>daviid: I'm a postgres committer but not on the (small) core team
<janneke>oh, my fixnum-size addition to the assembler is /almost/ correct
<janneke>now looking into a problem with the `-1' constant
<pinoaffe>I need to do some tree matching and I don't think there is a pre-made matcher that does what I need, does anyone have some pointers to writing tree matchers, or a relatively simple one to use as a reference?
<pinoaffe>I want to do partially ordered matching so that '(a 1 2 (b 3) (c 4)) and '(a 1 2 (c 4) (b 3)) are indistinguishable for the matcher, while '(a 2 1 (b 3) (c 4)) should be different, and I wasn't able to find something like that
<chrislck>pinoaffe: see (ice-9 match)
<chrislck>or maybe not
<chrislck>what's the exact matching rule?
<pinoaffe>chrislck: it doesn't do the unordered matching I need, but I'll take a look at the initial simple-match implementation and look whether I can modify it
***jonsger1 is now known as jonsger
<leoprikler>so atoms should have the same order but tagged pairs should not?
<leoprikler>I would suggest sorting the list in advance with this in mind, using a predicate that puts atoms before lists and lists before other lists à la symbol<=
<janneke>hmm, it seems to bo combination of scm_less_p and -1 (and .go compilation)
<dsmith-work>{appropriate time} Greetings, Guilers
<dsmith-work>RhodiumToad: Ever consider adding Guile to Postgres?
<mwette>pinoaffe: check Wingo's tree-fold article: http://wingolog.org/pub/fold-and-xml-transformation.pdf
<mwette>pinoaffe: source code to sxml-match in the module (sxml match) may provide some help
<mwette>in sxml-match attribute order does not matter
<mwette>pinoaffe: sorry, not clear if you are looking at trees or lists with only atoms and pairs; what's the form of your input?
<RhodiumToad>dsmith-work: I believe someone did a pl/scheme a while ago, not sure what implementation they used
<RhodiumToad>dsmith-work: it should be doable with guile but _only_ if built without threading
<RhodiumToad>(the postgres backend is possibly one of the most threading-hostile C environments in existence)
<dsmith-work>heh
<dsmith-work>RhodiumToad: I had considered it a long while ago. (Guile 1.3 or 1.4 I think). I was working on an Apache guile plugin and I thought it would be really cool to have Guile running in the browser, web server, AND the database.
<dsmith-work>Back then, guile had no dependencies. Bignums, GC, and Threads were all in libguile.
<RhodiumToad>bignums and gc shouldn't be a problem
<guix-vits>whoami
<guix-vits>cool
<guix-vits>Hello people, hello str1ngs. I've found out that the fan on my RockPro is actually working: `echo <number> > /sys/class/hwmon/hwmon<number>/pwm1` makes it. So fan is good.
<str1ngs>guix-vits: nice, I just ordered one yesterday.
<chrislck>sneek: botsnack
<sneek>:)
<chrislck>good bot
<str1ngs>guix-vits: how well is guix supported can you make system image with guix?
<guix-vits>str1ngs: IDK. Those SBC so new to me, that i'd just installed Armbian. What do You mean?
<str1ngs>guix-vits: I was just wondering if you can create a rock64 system image using guix build system ./rock64.scm
<str1ngs>I run guix foreign on my pinebook, with debian buster.
<guix-vits>str1ngs: I think _i_ can't, but ..:)
<guix-vits>str1ngs: is suspend works?
<str1ngs>hmmm I don't use suspend on my pinebook but let me check for you.
<str1ngs>I have a nvme so normally I just power off.
*janneke has a pretty nice working patch for x86_64-mingw
<janneke>found another void*/long/fixnum confusion thing in vm-engine.c
<janneke>sizeof long!=sizeof * -- talk about unwise platform choices...
*str1ngs all hail janneke! The slayer of bugs!
<dsmith-work>sneek: seen wingo
<sneek>I think I remember wingo in #guile one month and 20 days ago, saying: all hail godbolt.
<dsmith-work>heh
<drakonis>one true king godbolt
<dsmith-work>I wonder what the latest changes wingo has been working on are about. Jumptables? As in cond/case ?
<guix-vits>gb
<daviid>RhodiumToad: nice! and good to know as well :) one always needs a postres expert they can quizz if needed :):)
<daviid>*postgres