IRC channel logs


back to list of logs

<Pastaf>well, for whatever reason it's trying to download everything again when I remove the internet connection
<Pastaf>and my choice is monitor or internet connection, since I don't have any extra wifi hardware dongles that don't require nonfree firmware, and I can't use either already in use
<ngz>You may want to ding into guix-devel ML archives. I recall someone explained how to copy a whole store to another machine, which is really what you're after, AFAIU.
<ngz>to dig*
<koz_>Has anyone tried to put GuixSD on a BeagleBone Black? If so, how did it go and how could I do it?
<efraim>will it run the linux libre kernel?
<koz_>efraim: Yes, it should.
<koz_>(I think people got Trisquel and Parabola kickin on it)
<koz_>I don't actually need it to run X or Wifi at all.
<efraim>easiest way would be to run guix on top of something else and then run guix system reconfigure on it
<efraim>but i dont think our linux libre kernel has a definition for arm
<koz_>So would I have to write my own?
<koz_>I mean, I guess I could do Debian/Guix, but I'd like to get GuixSD running on it if possible.
<efraim>or see what trisquel/parabola do
<koz_>efraim: Alrighty, will do.
<efraim>if you run guix system reconfigure it'll install guixsd over the current running system without wiping it
<koz_>efraim: Really? It can just do that now?
<koz_>(or could it always?)
<efraim>it could always, it doesn't need most of the FHS directories
<koz_>I see.
<koz_>So basically, I should go something like easy-to-setup-OS -> Guix -> 'guix system reconfigure'?
<efraim>that's what I'd do
<koz_>efraim: OK, will do that then.
<timotheusbooth>hey, i am trying to install guixsd and the system init command failed at grub installation because i didnt have a bios boot partition, i have created one now. how do i go about continuing installation presuming i can without starting all over again
<koz_>Dunno if this is the right place to report this, but the 'Packages' page of the GuixSD site is causing my browser to lag and/or freeze.
<OriansJ>koz_: that is because the number of packages that GuixSD has is growing quite large and no one has gotten around to fixing that UI issue.
<koz_>OriansJ: I see. Well, I'd help, but I'm a complete Web noob.
<OriansJ>koz_: We all help in ways that we think matter. You never need to ask to help, just grab the shovel you think you can do the most good at and share what you've done with the group.
<koz_>OriansJ: Oh, I know - I'm currently doing some LibreJS-compliance-related stuff for another site. However, I just don't even know *how* I would go about solving that issue.
<OriansJ>koz_: well, experimenting with a couple crazy ideas might provide you some insight to solving it. But like all problems it is about amount of problem space exploration. Sometimes the solution comes from someplace completely unexpected
<koz_>OriansJ: Yeah, but it'd involve learning a lot of Web things, which I really don't have the time or headspace for. Being a full-time PhD student is pretty darn draining.
<OriansJ>koz_: what is your PhD in?
<koz_>OriansJ: It's in some intersection of logic and distributed systems.
<OriansJ>koz_: sounds like it isn't much of a passion. What are you actually passionate about doing?
<koz_>OriansJ: Oh, I'm *very* passionate about it. I just don't wanna spend a lot of time explaining it in a lot of detail (although I can!), because I've learned people tend not to want walls of text.
<koz_>If you want a wall of text, I'm happy to provide you with one.
<koz_>My choice of 'some' there was more to do with 'hard to define' not 'hard to give any fucks about' - though on a second read, I can see why it might have confused you.
<OriansJ>koz_: go for it; tell me where it is amazing.
<koz_>OriansJ: Basically, there's a pretty neat problem called 'logic synthesis'. Basically, given a truth table, determine the shortest (for some definition of 'shortest') propositional logic formula whose truth table is the same as the given one.
<koz_>Pretty important for circuit design, but also nicely generalizes a lot of constraint-related problems.
<koz_>My idea is to solve it using a distributed system (specifically a mesh network). Right now, I'm working through a few ideas of how to do this while avoiding running face-first into the *very* large global state dependency this problem tends to induce.
<koz_>I've thought of a way to do it, but I'm currently seeing if it can give satisfactory answers. Additionally, because my approach is basically evolutionary in nature, analyzing it is *hard*.
<koz_>What I really wanna do is get a distributed solution to logic synthesis that can run on a very cheaply-built meshnet (like, made of RaspPis or something similar) and still attack large instances of that problem effectively.
<koz_>Right now, I've hit a bit of a frustrating low, but it's more to do with my somewhat-inadequate programming skills. :(
<koz_>That's the *slightly* more wall-of-text-y solution.
<OriansJ>koz_: have you considered the genetic fitness optimization problem in biology?
<koz_>OriansJ: I haven't actually. Mind linking me?
<OriansJ>Image a cell culture, every cell has an absolute fitness and everytime it mutates that value can change in massive ways
<OriansJ>However its fitness is also relative to the cells around it
<OriansJ>Yet novelty first search has shown us that directed optimization tends to get stuck on local optima
<OriansJ>You are familiar with novelty first search right koz_ ?
<koz_>Maybe not by that name. Could you describe what you mean by that?
<koz_>OriansJ: Thanks - I'll read that!
<koz_>Are you a PhD student/PhD holder too?
<OriansJ>In short novelty first search, is not about getting to any place in particular in a search space but reducing the total path to as many places as possible.
<koz_>OriansJ: This is actually really interesting - my supervisor posed questions to me regarding very similar issues to the ones this paper seems to discuss just this week.
<koz_>Thank you for linking me to this - I will *definitely* read it!
<OriansJ>koz_: I am alway glad to help the community.
<koz_>OriansJ: Are you a researcher as well?
<OriansJ>when my other work permits.
<koz_>OriansJ: What *is* your other work, out of interest?
<OriansJ>Currently, I build software to help people obtain access welfare.
<koz_>I'm guessing it's nonfree software?
<Sleep_Walker>sorry if it is known problem but have you seen this?:
<Sleep_Walker>suspicious ownership or permission on `/gnu/store/3a7xa0khsriwgcmw5vk90d8dv8h2hbfn-glib-2.48.0-bin'; rejecting this build output
<OriansJ>Well I am working on reducing that koz_
<koz_>OriansJ: That's great!
<koz_>Any freedom in software which wasn't there before is a victory.
<OriansJ>Prior to that, I was a leading researcher in artificial intelligence but I didn't want to work in the direction where there was funding [drones/HFT] and left the field.
<koz_>I see. Sorry to hear that you had to leave research behind due to funding shortages. :(
<koz_>(or rather, lack of funding in things that interest you
<ifur>can the git-fetch method be used for svn?
<OriansJ>AI winters are inevitable because hard problems in AI seem so simple.
<koz_>OriansJ: Yeah - I remember the big promises made in early AI and how hard *they* proved to be.
<OriansJ>ifur: If you use a git-svn bridge
<Sleep_Walker>drwxrwxrwx 3 guix-builder1 guixbuild 4096 29. kvě 11.55 /gnu/store/3a7xa0khsriwgcmw5vk90d8dv8h2hbfn-glib-2.48.0-bin
<koz_>(well, as in, I know of them)
<koz_>And the hard problems are of course the most fun.
<OriansJ>koz_: You have to remember it only took them 3days to write a program that could solve every math problem given on every standardized math test available to them.
<koz_>When/where did *this* happen?
<ifur>OriansJ: well, AI has a lot more real world applications these days in other fields
<ifur>so interdisiplinary applications of AI are definitely there
<ifur>studied cognitive sciences a while back, and quite frankly, the amount of bullshit some people peddle is discouraging... philosophy of mind in general is a joke
<ifur>it gives off a better impression than it should because those that you do hear about aren't everage -- but should be.
<OriansJ>koz_: MIT 1959 in exactly 47 lines of lisp
<koz_>OriansJ: Woah, neat.
<OriansJ>ifur: AI is a fuzzing tool for mapping out a problem space. When the problem becomes solved, you don't bother to use AI as it is inefficient and makes hard to solve bugs
<OriansJ>koz_: It ultimately grew and added features and became maxima
<ifur>OriansJ: i wasn't refering to AI as in search algorithms, but intelligent system; especially self-improving ones
<koz_>OriansJ: That's really cool.
<ifur>of which we are barely touching upon these days
<ifur>academics tend to speak of AI and mean search, lay persons think that equates to intelligence and conscious -- why else would researches speak of AI?
<ifur>in many cases its stretching it quite far to even begin calling it intelligent
<OriansJ>ifur: The problem with General purpose AI is how to quantify optimization. Aka, how not to make an AI that converts all of our atoms to paperclips.
<ifur>OriansJ: which is a definition problem that makes philosophers cringe
<OriansJ>koz_: This is a wonderful example of something everyone though was a massive hard problem that they solved really really quickly
<ifur>AI as a question is ill-defined, computer scientists have no business speaking of intelligence no matter how advances the search is when the former is not defined, understood or even posed as a proper question
<ifur>you cant make an evolutionary algorithm and call it intelligence for example
<ifur>I admit, that the terminology is so muddled now that its meaningless, but the invention of strong AI was so that people would be able to continue bullshitting about the body and soul problem
<OriansJ>ifur: Intelligence is poorly defined, because we don't even know if humans can be classified as intelligent.
<ifur>and now computer scientists latches on and say, no no, AI is not the same as STRONG AI.
<ifur>nono, we aren't talking about general intelligence, because that would be strong!
<ifur>here we speak of weak AI which means search algorithms, which has nothing to do with intelligence
<OriansJ>ifur: All language is fuzzy because no 2 people mean the exact same thing, even when talking about the exact same thing and using the exact same words.
<ifur>brilliant! Wittgenstein would be impressed by this language game
<ifur>define your premiss, dont hinge on others lack of understanding of the domain specific use of a term.
<OriansJ>ifur: Now if we assume humans are an intelligence. We then can assume some of the things that humans have or do are essential to any intelligence.
<ifur>actually, you can't assume that
<OriansJ>ifur: You can assume any arbitrary thing you want.
<ifur>you need to be specific
<ifur>sure, if we are talking abtract maths or physics
<ifur>but I can counter argue by saying that farting is not intelligent
<ifur>and then you can start arguing that taking a shit is intelligent given the fact that there are roughly 100k neurons in human intestines
<OriansJ>ifur: why do you think there is a difference between a distinct abstract math and the universe in which we exist?
<ifur>this sort of discussion makes no sense, you can't arbritrarily assume anything when definined a term and its meaning
<ifur>because even if meta-physics necessitiates something being possible
<OriansJ>ifur: Yes you can because all language is simply the collective results of multiple shared assumptions.
<ifur>you also need to argue for THIS universe, WITHIN our perspective of it
<ifur>otherwise its meaningless
<ifur>you cannot magically go from an ontological category and assume that there is an epistemology there
<ifur>that form of argument has gone down in history as the "ontological argument for god"
<ifur>the validyity of that sort of reasoning died in the 19th century
<ifur>or was it 18th? cant remember :/
<OriansJ>ifur: Failing tests are the only way to weed out that crap
<ifur>im not saying AI is bullshit, I like it... the confusing use of terminology annoys me tho
<ifur>in many cases i would prefer adaptive system, rather than intelligent system etc
<OriansJ>ifur: AI has become a bullshit marketing term but the real working parts still work where they are applicable.
<ifur>bingo! applies maths and discrete maths should have more funding
<ifur>the fact that researches need to use buzzwords to market their research to get funding is terrifying
<ifur>scientists are rapidly losing the ability to effectively communicate accross scientific fields because of this
<OriansJ>ifur: The reason is because even though everyone agrees that good things need to be funded more and bad things need to be punished more, we all have different assumptions and what we mean about those terms are inevitably different.
<ifur>well, philosophy matters also :S
<ifur>thats usually where terms are defined and possible meanings are studied
<OriansJ>ifur: philosophy is a useful and interesting tool but sadly like all tools there are many places where it just doesn't belong.
<ifur>i dropped out of philosopy to do computing for physics, FYI :)
<ifur>well, philosophy today is actually the study of the history of ideas
<OriansJ>ifur: Is it because you enjoy the computing or you wanted to know more about the physics?
<ifur>philosophy has become and oxymoron and is quite impotent today in the scientific sense
<ifur>i would never stand a chance at a phd program in philosophy anyways
<ifur>i argue too much, and that isnt popular
<janneke>ifur: what about eg, David Chalmers?
<OriansJ>ifur: no one ever said being popular is the same as being right.
<janneke>ACTION has the feeling he isn't dead yet
<ifur>janneke: the guy that talks about the soul-mind problem and hasn't been able to move on from Descartes?
<ifur>OriansJ: i can read and think without participating in academic philosophy
<OriansJ>ifur: Also, one must notice the length of any argument is inversely proporitional to the number of objective facts available.
<ifur>OriansJ: in philosophy you can studied the space of possible semantic interpretations and wether it would break laws of logic or meta-physics
<ifur>and by doing so actually find insights that makes you ask different questions, or at least ask them in a better way
<janneke>ifur: yes, one of the few who dares to look into what 'I am' means
<janneke>ACTION starts to understand even more how daring that is
<ifur>janneke: read nietzsche instead
<ifur>chalmers is a poor existentialist
<ifur>even sartre is better
<ifur>but wittgenstein, heidegger nietzsche and kierkegaard is more fruitful
<ifur>chalmers is only interesting if you care about religion
<ifur>no need to circumvent the church anymore to be allowed to talk about these problems
<janneke>i'm sorry you don't care for the hard problem
<ifur>and woith that, the mind body problem goes away, and you can start focusing on what really matters, namely what is consciousness; and in meta-physical terms can the soul be said to fall into a category
<ifur>mind body problem is not the hard problem, that problem only exists if you take meta-physics in a litterall sense
<ifur>which means you actually do believe in heaven and hell in the litteral sense
<ifur>most rhinkers through history have however focused on the meta-physics, e.g. the laws governing both heaven and hell (space and earth)
<ifur>its all language games :(
<janneke>i still like people who dare look at the question [why] are we conscious
<ifur>that isnt daring, orginal or new
<ifur>its a series of arguments that's been recycled since the ancient times
<ifur>more daring is to ask if consciousness is just an illusion
<ifur>and how many forms there may be
<janneke>see how daring it is: try getting a paper published if people dismiss your topic
<ifur>why isnt a meaningfull question without the how
<ifur>you can never ignore either epistemology and ontology; which are the two branches of philosophy, full stop.
<ifur>you never find insight if you do, you just end up circling back onto the question
<ifur>perhaps using different words, or muddeling definitions along the way
<ifur>this is why philosophy is dead
<ifur>people butting in believing its a game of oppinions, when the opposite is the case
<jlicht>ifur: well, that's just your opinion man ;-)
<ifur>dilletants beliving that if they "behave" like a philosopher, or immtate their "style" they will become one.
<ifur>historically the study of philosopy has been a litterary one
<jlicht>does guix have any special requirements for packaged fonts? Are pre-made otf files acceptable?
<ifur>jlicht: don't know, you could git grep the repo and see if some exists?
<jlicht>ifur: seeing as this seems to already be happening in several font packages, I guess it is a sane assumption that this is not an issue
<OriansJ>jlicht: well I am not certain but I am guessing that the general idea is if possible make the source material and any build instruction required to produce the end result. As I am unfamiliar with how difficult it would be to modify an otf to suit me, ideally the prefered form is one in which it is easiest to modify.
<OriansJ>Thus if I had to guess, treat it like javascript. It is fine to have the compiled result, as long as we have the original source and the build instructions that produces that compiled result.
<OriansJ>Or I could be entirely wrong and it would be like Images, in which the source material is not required as the resulting formats can be as easily manipulated by all as the source material itself.
<jlicht>OriansJ: That was what I was having doubts about as well. Of course it would be nice to have the actual 'source' files, but looking at it as being images, this does not really make sense
<jlicht>oh well, I'll keep it in my own tree for now and send it to the ML once I figure out how to build the fonts myself.
<OriansJ>jlicht: I'd love to see when you finally figure the build process out. :D
<jlicht>OriansJ: It seems to involve AFDKO, which seems to be free software at first glance, but has a very... interesting set of build instructions xD. It contains a custom version of python and other such niceties \\o/
<avph>hi, I wanted try guixsd in a qemu vm, but on 256M and 512M memory it failes to install because it runs out of memory (while compile emacs IRC). What amount of ram do you recommended?
<lumidragon>after starting a container using the script produced by 'guix system container' how do login or gain access to the container?
<lumidragon>and hi ^^
<lumidragon>avph: oh and avph you problem sounds similar to mine. Did you mount any swap?
<avph>nope, should I?
<lumidragon>I had to do that when I did my first install.
<lumidragon>I some created for the new install, but the compile failed part way.
<lumidragon>when I turned the swap on for the install process everything went ok.
<avph>ok, I will try with swap
<lfam>Some packages require a huge amount of RAM to compile
<lfam>lumidragon: Does your container creation command at least exit successfully? I've never used it so I tried it, and it crashes
<lumidragon>thing is it doesn't exit. it starts sheperherd and bunch of stuff.
<lfam>lumidragon: In that, I think it's running for you
<lfam>In that case*
<lumidragon>I figured, since it asked for that long seed lol, just trying to figure out how to get access.
<lfam>lumidragon: If you don't need SSH access to the container, you can remove the lsh-service and then you won't have to dance on the keyboard. The random seeding is required by lsh-service
<lumidragon>oh kk, noted. I got past that part.
<lumidragon>how do I login into it?
<lfam>lumidragon: I still can't get the container creation to work for me, so I don't know
<lfam>Are you using `guix system container` or `guix container`?
<lumidragon>guix system container atm.
<lfam>You want the system variety, probably
<lfam>Does the command that creates the container print the name of a file to run?
<lumidragon>it gave me script, and that's what I ran.
<lfam>You don't get a login prompt?
<lumidragon>although it said failed to start tty
<lumidragon>and I started the script using sudo
<lfam>"failed to start tty" that doesn't sound good!
<lfam>I wonder if something is wrong with the container code, or if your OS declaration has some problem?
<lumidragon>I used the same config that my current system is using.
<lumidragon>prb why I forgot about the lsh thing.
<lfam>Too bad one of the container experts isn't here right now
<lfam>I haven't really used that feature much yet
<lumidragon>bummer, if I don't make any progress I'll just send a query to help-guix.
<lfam>It's worth an email to help-guix or bug-guix if you are really stuck
<lumidragon>kk was thinking the same thing thanks :)
<lfam>These tools should *just work*
<lumidragon>pretty much what one would expect.
<lumidragon>I was curious what it can do, as was trying out rkt in another vm as well.
<lumidragon>oh ok figured it out.
<lumidragon>turns out I was using the wrong pid for the "guix container exec" command
<lumidragon>I needed to get a child process of the container.
<lumidragon>another question, is there a way to keep tmp guix-build-* directory around after a sucessful build?
<efraim>if I want it to stop somewhere specific I add to the phases in arguments (add-after 'build 'fail #f)
<efraim>or after configure sometimes
<lumidragon>oh ok thanks, that will very useful.
<ng0>when did I file the lispf4 bug?
<ng0>101 days ago.. hm
<ng0>so not after the last fixes of the lispf4 repo
<lfam>Does anyone know I can get a patch of this SVN revision:
<ng0>could you check out the revision, check out the one which the current base is and create a diff between them?
<ng0>", and"
<lfam>ng0: Surely, but I don't know how to use SVN
<lfam>I'm used to web interfaces to Git that helpfully offer the diff to you
<ng0>me neither, but you can use git-svn
<lfam>I guess that's worth a try
<ng0>though for svn there's also help out there, I'm not fluent enough in it to recall it
<lfam>Yes, I'm being lazy ;)
<ng0>i checked out guix repo on my server for stagit, the log pages takes really long to load
<ng0>*log page
<lfam>I bet they load faster the 2nd time, once the OS has them cached
<ng0>answer 10 could help
<ng0>but you still need to checkout the revisions.
<lfam>I'm trying git:svn
***frafra_ is now known as frafra
<lfam>Anyways, while I sloooowly clone the pcre2 SVN repo, there are plenty of other bugs reported by `guix lint -c cve`. Help wanted
<ng0>did you run the checkout without a version and fetch all the history? oh :/
<lfam>Yes, I just did the naive clone. It's done now
<roelj>This is the second time I run into an issue with libgcrypt. When trying to run guix-daemon, it ends with: "error: libgcrypt version mismatch".
<roelj>Has anoyone else had this? And if so, how did you fix it?
<civodul>lfam: congrats on fixing all the old CVEs!
<civodul>roelj: it means that the libgcrypt version found at compile time does not match that found at run time
<civodul>so, 'ldd', 'strace', and all that :-)
<roelj>civodul: Alright. Thanks :)
<lfam>civodul: The list is not done yet!