IRC channel logs

<RISCi_ATOM>Hello everyone o/

<grasshopprWhoppr>hi

<xisiqomelir>good morning friends

<tadni>xisiqomelir: Good evening.

<xisiqomelir_>Has anyone virtualized 0.7 successfully in virtualbox?

<xisiqomelir_>When I invoke "guix system init" hydra.gnu.org is unresponsive

<xisiqomelir_>I can ping it from my host though

<xisiqomelir_>Ok it works with '--no-substitutes'

<RISCi_ATOM>xisiqomelir: I have in qemu and on physical hardware ;)

<tadni>2015 will be the year of the Guile Desktop!

<xisiqomelir>Ya I have it working in vbox now

<xisiqomelir>I'd be happy to switch to this as my primary desktop with a few more packages

<xisiqomelir>and maybe a guix front-end in ncurses or something

<tadni>xisiqomelir: I'm fine with a textual installer, but eventually we want a psuedo-graphical to graphical installer for the every-person non-technical user.

<xisiqomelir>I think the debian installer can be appropriated/repurposed quite straightforwardly

<tadni>xisiqomelir: Really, I just need a simple way to enable guile-wm as my main graphical session and set up wifi easily are my two big caveats. That and I need an .iso for my laptop.

<tadni>My main laptop.

<xisiqomelir>for installing from optical media?

<tadni>xisiqomelir: Yeah, my bios don't seem to support usb drives.

<xisiqomelir>What is your laptop?

<tadni>xisiqomelir: A newish samsung with UEFI.

<xisiqomelir>Can you disable secureboot?

<tadni>xisiqomelir: Yeah, I can.

<tadni>Even has an emulation layer for older bios.

<tadni>I wouldn't think you could easily repurpose Debian's installer.

<xisiqomelir>Have you tried another usb key?

<tadni>xisiqomelir: You know, I might give it another try... but lack of GUI is a big thing for me.

*tadni` attempts to dd onto a new usb drive.

<tadni>So is guix 0.7's tarball actually 0.6? I downloaded it to my Fedora box and it says 0.6

<grasshopprWhoppr>Where do you see that?

<tadni>grasshopprWhoppr: I tried downloading Guix, from guix and it's 0.6. I pulled as root and am now building 0.7,

<tadni>Which I need to because it wasn't in /gnu/store so I can't authorize the repo yet for substitutes.

<tadni>Okay, Guix on my Fedora install is a go. :^)

<svetlana>tadni: hi.

<svetlana>tadni: how much space does a guix install take?

<tadni>svetlana: Like on an already extant distro?

*tadni needs to figure out how to generate a vm-image.

<svetlana>tadni: I'm looking for a usb drive to put it on. need to know the size

<tadni>svetlana: So, the installer for the distro then?

<svetlana>I would install it onto a usb flash drive if at all possible

<tadni>Nearly 800mb for the Guix installer distro.

<svetlana>what would that do? function like an installer and/or a live cd with a need of another separate partition if I'd like to install it?

<tadni>svetlana: You can do a little with the installer image in-regards to playing with Guix, but it's more-so specific to actually installing to an actual device.

<tadni>I would maybe carry a vm-image with you on a flash drive.

<svetlana>I'm just trying to figure out how to get it run on the computer itself without another host os. from your answer it looks like I have to install it to try it properly

*tadni isn't sure what to put for the "-s" flag for "guix system vm-image". :^P

<tadni>svetlana: Running without a host OS, certainly implies you want to use it as the main OS on that designated box... or am I still misunderstanding you?

<svetlana>to me, vm-image sounds like running it within a host os

<svetlana>I can triple boot if there is no live cd. if it works, I don't mind using it daily, but I'd rather leave other two OSs working too

<tadni>svetlana: You need a host os yeah, but the store is separate from said OS.

<tadni>svetlana: I'm not sure if the installer is at a point where you can easily dual-boot other stuff with it.

<svetlana>hm

<svetlana>as a workaround I could install from one flash drive onto another one? :s

<sjamaan>It's possible to omit grub iirc what the manual says

<sjamaan>So you could make it dual-boot by configuring another distro's grub settings manually

<svetlana>I already have grub installed here

<svetlana>yes

<svetlana>ok, so you told me that the 0.7 'image' is an image of an installer, not an os. I need to think where I'll install it /to/. what does installer do about this? is it happy if I point it to an existing partition, or an existing usb flash drive?

<tadni>Also, unless you are ready to go without a GUI and without Wireless, without a fair amount of work -- I wouldn't use the install image for another release or so.

<tadni>svetlana: You could probably manage that.

<sjamaan>Check the manual: it explains how to use gparted to create the partition, then you mkfs it, mount it and install to there

<svetlana>ok

<sjamaan>tadni`: I think there should be no problem if you can use ethernet, right?

<svetlana>I don't have ethernet access to the web, only wifi

<svetlana>I'll do some more reading as I managed to miss the information you gave. thanks.

<sjamaan>That would be problematic, I think

<sjamaan>Unless the tools on the install drive are sufficient to set up wifi, but I don't know about that

<sjamaan>(I've only tried it in a VM)

<svetlana>(I can tether it to a mobile with working wifi, but I would have to learn to do that by hand)

<svetlana>wpa sapplicant and /etc/network/interfaces need to work, that's sort of enough iirc

<sjamaan>Yeah, should be

<sjamaan>Though I always end up fighting with wpa_supplicant

<sjamaan>It's just impossibly difficult at times

<sjamaan>wicd is much simpler to use (with wicd-curses)

<tadni`>Sorry.

<tadni`>sjamaan: Nope, just "dhclient eth0"

<tadni`>svetlana: Ludo (I think I never got a conformation in this, but it appears to be the case) accidentally shipped guix 0.6 in the 0.7 install image. You'll need to "guix pull" as root and "guix package -i guix" then to grab the latest, but wpa-supplicant and wireless-tools are in the latest release.

<tadni`>davexunit has been playing with it and you can probably talk to him tomorrow if he was sucessful.

<tadni`>I'm not sure of the Gui situation, besides it is more-or-less hardcoded to use windowmaker at this point.

<tadni`>Yah, Alpha-tier software! :^)

<sjamaan>windowmaker!

*sjamaan has fond memories of using that on university Solaris boxes, where everything else was crap

<tadni`>By the end of the year (probably next release or-so at this rate) we'll have something solid enough that a fair amount of higher-level technical users can use without much of a worry.

<sjamaan>I think it's impressive how quickly Guix has matured already

<tadni`>This is not something I'd give anyone in my family, probably for another year or-so ... but, the future is bright!

<tadni`>sjamaan: Ludo is a powerhouse.

<svetlana>ah.

<tadni`>Man with a vision and a passion can do a lot.

<sjamaan>yeah

*tadni` wishes he could be of more general help ... I'm not technically knowledgeable to really to anything too stellar. I packaged slim and it's depends awhile back ... that's about it.

<svetlana>anyway, I need to pipe the installer onto a flash drive; read on wpa_sapplicant and/or steal my existing config; and add a partition to install it to. how much does it need?

<sjamaan>Is there a list of things to be done somewhere?

<tadni`>sjamaan: As in what needs to be done, for the distro to considered stable -- or?

<sjamaan>yeah

<tadni`>svetlana: I'd say 4 gigs minimum if you want to install anything above the base-system. I don't know the number off hand, nor know how to calculate it. I guess I could look how big my store is ... all I have in it besides the base-system is Emacs ... but the installer has a fair amount of stuff not in the base-install of guix on a host system.

<tadni`>sjamaan: Not to my knowledge, though such a thing might be a good idea. Really once we have a user base of about 100, I think a focus should be to establish some sort of community board to make decisions like what they have in Debian or the like.

<tadni`>Right now you have to classes of users you want to satisfy, the technical user and the layman both of which have very different needs. The technical user will be the one you need/should cover first, because they will be the one's supporting your platform for the most part.

<tadni`>This is why, while a psuedo-to-actually-graphical installer, while I think is important, should not be a prime focus now.

<tadni`>Or in the "near" future.

<sjamaan>Will Guix be targeted at the complete beginner (like, say, Ubuntu or Elementary OS)?

<sjamaan>I think it's important to have some sort of scope or target audience

<tadni`>Ideally, I'd like to see GNU be more-or-less the development tool of the GNOME project ... I think we may be to late on this front, but I think such an audience is really important in the foss community.

<sjamaan>You'll get pulled apart by competing forces if you target technical people _and_ beginners, I think

<tadni`>sjamaan: We'll probably have a core distro and "spins" like Fedora or Ubuntu, especially seeing how easy it is to derive a new guix based system.

<sjamaan>Technical people prefer to understand their system and have it be "simple" in that sense

<sjamaan>Beginners would tend to prefer things to just be "intuitive" (whatever that means)

<svetlana>I agree. I am asking just enough to figure things out; not demanding you to waste time on an installer. The relevant things are (or will be) in the documentation anyway. (I heard that being graphical does not necessarily mean being easier to comprehend.) :)

<tadni`>It's not up to me though, right now the direction is mostly in the hands of Ludo and hopefully at some point a group body.

<Steap>sjamaan: SInce we won't package non-free crap, I think it won't be suitable for real beginners :)

<sjamaan>Steap: That's a bit defeatist, don't you think? :)

<tadni`>Steap: Really, the only non-free tool I now a lot of people won't give up is Skype. I'm hoping Tox really catches on in this regard.

<sjamaan>tadni`: Don't forget drivers

<svetlana>gnome is so much different from windowmaker that I can't imagine GUIX a development tool for it until (i) you're doing that from within gnome, or (ii) these two things become integrated better.

<Steap>tadni`: Skype, Flash ? :p

<sjamaan>The only non-free thing I had to install on my Debian laptop at home was a wifi driver, iirc

<Steap>+ drivers, indeed

<sjamaan>Flash is rapidly becoming irrelevant

<Steap>real users use non-free stuff

<Steap>sjamaan: not to the average Joe :)

<Steap>Guix is mostly targeted at Free Software enthusiasts

<Steap>imho

<tadni`>sjamaan: Drivers become less and less of an issue each year though. Not saying it will go away, but when I started using GNU+Linux like 5-6 years ago ... it was near impossible it seemed to have wireless drivers working out of the box, 3 laptops later and never had a problem since.

<svetlana>I don't think we need to worry about people using nonfree things -- drivers are being added in, there is gnash, and people can live without skype anyway

<Steap>tadni`: you still have a lot of wifi cards that require a non-free firmware

<tadni`>Steap: True, but it's a trend that appears to be getting better and I hope will continue to.

<sjamaan>Until the next new technology comes along ;(

<tadni`>2015-16 will be a very solid new beginning to the GNU System I think. Both GNOME3 and Guix will be pretty mature, so will Tox.

<tadni`>It's a very promising future to a FOSS advocate, now we just need to get better at selling ourselves and our purpose... which is arguably harder.

<sjamaan>Convincing people that software freedom is important seems to be extremely difficult

<sjamaan>They want shiny features and "access to the code" (or even asking someone else to change it for you) is too technical

<tadni`>sjamaan: I think a fair amount of the difficulty is sadly that our software, from the UI perspective has been lacking until recently. It's sad that it matters to so many people ... but it's an important factor.

<sjamaan>I'm not sure that's sad

<tadni`>I think the FSF should really invest in teaching materials. Empowering people to learn how to code (and I think that this should be a mandatory part of public education going onward) I think will convince a lot more people that such things are important.

<tadni`>sjamaan: People are not willing to make compromises for freedom. It's understandable, but sad... both for them not willing to do such things, but too that we've been lacking so much until very recently.

<tadni`>GNOME3.x is really the first Free Software Desktop I've put non-technical people in front of and not actively instruct them what to do.

<sjamaan>The danger in that is that when we can create something that's as easy to use as the proprietary programs they're using now, they might switch back to using proprietary software as soon as something easier to use comes out

<sjamaan>Trying to win people over by competing purely on features and ease of use is the wrong battle

<Steap>You do realize everybody uses VLC because VLC just kicks ass, not because it's Free ? :p

<sjamaan>Yeah

<tadni`>sjamaan: Yeah, but the idea is that we actively state that freedom is the primary matter -- but lure people in by showing that the don't have to make compromises in-terms of usability (the scary bit).

<sjamaan>Steap: But those VLC users will switch back as soon as something easier to use comes out, or with better features. See also how many people switched from Firefox to Google Chrome

<tadni`>"Linux" has such a negative association to it for being completely textual and scary and it's a stigma that is going to be hard to lose.

<Steap>sjamaan: yes

<Steap>but it's a lost cause to get people to switch because of Freedom

<Steap>give them shiny features

<Steap>give them good software that's easy to use

<tadni`>Steap: Give them shinny features, but also advertise the importance of freedom while doing so. Telling people about Window's backdoors is also a good thing and assuring such a thing won't happen on "GNU".

<Steap>sure

<tadni`>There is a bit of fear mongering, but because the boggyman really is out there.

<Steap>but you can't go and tell them "hey that's Free, use it !"

<Steap>that only works with nerds like you and I :)

<tadni`>Steap: Which really is a bit sad, when one gives that a monochrome of thought too.

<Steap>not really

<Steap>there are lots of causes to support in the world

<Steap>and you don't support them all

*tadni` haz all da cauzezs

<Steap>heh

<Steap>e

<tadni`>Not all causes are directly relevant to me. Using a computer that has non-free software is a thing that I am effected by everyday.

<tadni`>I may not associate such a thing as a negative, but it is directly in my bubble. :^P

<xisiqomelir>Hey people are chatting!

<xisiqomelir>I'm going to be building openbox and sakura and midori

<xisiqomelir>I eventually want to get MATE (my preferred DE)

<xisiqomelir>but first it will require GTK2/3 deps and other things

<Steap>long way to go :)

<tadni`> /me is back, just showered.

<tadni`>Yeah, mate is probably a fair ways away.

<tadni`>Really, besides conkeror (which via extensions to icecat I can get most of the way to its usability anyways) I'm pretty much golden.

<tadni`>Everything I use (mostly Emacs) is already packaged already, but guile-wm needs some work. Namely mode-line support.

<tadni`>xisiqomelir: Mate is just updated GNOME2.x branch right?

<xisiqomelir>yes tadni

<tadni`>jmd: o/

<xisiqomelir>they removed some of the "dark arts" however

<xisiqomelir>so no more libbonobo

<xisiqomelir>and other such weirdness

<xisiqomelir>and it's ported to GTK3 already

<xisiqomelir>so the backend is modern

<jmd>tadni`: o\\

<xisiqomelir>hi jmd

<jmd>Hello

<tadni`>WHAT ... CONKEROR IS PACKAGED?!

<tadni`>Oh, yes!

<svetlana>"guile-wm"

<svetlana>mmmmmmm

<grasshopprWhoppr>:)

<grasshopprWhoppr>very lispy

<sjamaan>Is that a new thing?

*sjamaan remembers scwm, which was also in Guile IIRC

<alezost>sjamaan: it's new: http://www.markwitmer.com/guile-xcb/guile-wm.html

<sjamaan>Just now that I thought I finally settled for i3 :)

<davexunit>morning #guix

<Mathnerd314>I hate #nixos. but #guix doesn't have enough development momentum. How to reconcile these two conflicting emotions?

<sjamaan>Add to the momentum by coding on guix, of course :)

<davexunit>Mathnerd314: not enough development momentum? We have many active developers.

<davexunit>Mathnerd314: is #nixos not friendly or something?

<Mathnerd314>it's friendly, there's just no momentum

<Mathnerd314>lots of developers * epsilon is still epsilon

<Mathnerd314>and if you compare your "many active developers" to debian's list then I don't think there's a comparison there either

<FracV>I think you know that's a logical fallacy

<FracV>That's like claiming Snowden isn't a big mountain because Everest exists

<sjamaan>This is starting to get bizarre :)

<FracV>You're setting an arbitrary line for what you consider to be good enough, a line which you can move at any time, and then complaining when others don't meet it.

<xisiqomelir>Can we go back to why he hates nix?

<xisiqomelir>Because I don't see why anyone should hate nix

<xisiqomelir>I just prefer lisp

<Mathnerd314>FracV: I believe that is the general reasoning of complaining, yes

<Mathnerd314>FracV: although it is not really "arbitrary"

<davexunit>Mathnerd314: debian is a project over 20 years old, guix is 3 years old, please give us some time to get the "momentum" you are looking for, and please help out if you can.

<sjamaan>davexunit: Earlier I asked if there's a todo list of sorts

<sjamaan>Is there?

<Steap>yeah really, either it's complete crap and you're free to use something else, or you can git clone and run vim :)

*Steap added a subtle troll

<Steap>sjamaan: see the "TODO" file and the "plans for 0.8" mail that civodul recently sent

<davexunit>sjamaan: yeah ^

<davexunit>and there's always "package more software"

<davexunit>that's very important.

<Mathnerd314>so there's http://lists.gnu.org/archive/html/guix-devel/2014-07/msg00297.html and http://git.savannah.gnu.org/cgit/guix.git/tree/TODO

<davexunit>ah, yes. forgot about the TODO file.

<Mathnerd314>I'm not seeing "get over 3000 developers" on that list

<davexunit>that's not a technical item.

<davexunit>if you build it, they will come.

<davexunit>the 0.7 release has already increased activity in this channel.

<FracV>Mathnerd314: You don't see Debian-scale and 3,000 developers as arbitrary requirements?

<Mathnerd314>davexunit: ok, but I'm not seeing anything that would obviously cause 3000 developers to start flooding in either. So what is this "it" you refer to?

<Mathnerd314>FracV: not really, if Guix is a distribution then it should compare itself to other distributions

<davexunit>Mathnerd314: it is guix. I'm too interested in continuing this conversation if you're going to be so negative.

<davexunit>we're a small group of volunteer developers, we contribute when we can on our own time, and we're trying to make good software and good freedom respecting GNU/Linux distribution.

<Steap>Mathnerd314: what do you expect out of this conversation ?

<Mathnerd314>Steap: some idea of what to do next month

<davexunit>Mathnerd314: packaging software that is important to you but missing from our repo is a constructive thing to do.

<Steap>Mathnerd314: well, package stuff, try out the Guix system, report and fix bugs

<Mathnerd314>davexunit: I tried that, then I waited 2 weeks and someone else packaged it for me

<FracV>I tend to pick a project at a time and spend a weekend on it

<FracV>I'd be interested in packaging up some software for Guix

<FracV>Although annoyingly I'm busy the next 2 weekends

<davexunit>Mathnerd314: well that's a fluke. there's plenty of software to package that others won't think of.

<Mathnerd314>davexunit: sure, there's lots of software, but all of the software I consider important has been packaged already.

<davexunit>ah, okay.

<davexunit>I can't give much guidance, but we need more DMD services.

<davexunit>one for the openssh server would be cool

<Mathnerd314>but I don't use openssh...

<FracV>I use openssh, and I'll most likely be using it on the machine I'll use Guix on

<FracV>I'll check in to make sure it's something that still needs doing when I'm a person with free time

<Mathnerd314>so maybe now you can see why from my perspective Guix isn't moving; it added the packages I needed, and then it stopped.

<davexunit>I don't understand. it hasn't stopped. there are still so many things to do to have a more usable system.

<Mathnerd314>oh, sorry, I meant NixOS. Guix is indeed different

<davexunit>oh.

<Mathnerd314>but not very different, AFAICT Guix is headed in the same direction as NixOS (minus some nonfree packages)

<Mathnerd314>so I guess it's traveling into a dead end

<davexunit>I don't know how you reach that conclusion.

<FracV>If it has all of the packages you want, then why is it a dead end?

<FracV>Surely what you're complaining about is a piece of software going from being alpha to being stable?

<Mathnerd314>because packages aren't static objects, and the Nix model treats them as static

<Mathnerd314>there's no room for change in "purely functional package management"

<davexunit>what do you mean as static?

<Mathnerd314>just that; they never change, they always have the same SHA256 hash, etc.

<davexunit>you can upgrade package descriptions easily.

<FracV>I don't know much about nix, but from what I *do* know I'm fairly certain the point is to use a purely functional language to handle dependencies really nicely, including for updating and rolling back versions of packages

<davexunit>Mathnerd314: I think you've missed the point.

<Mathnerd314>davexunit: not really, last time I tried (a few days ago) it took IIIgIIIIIIgII = 14 steps. "easy" is pushing a button = I = 1 step

<Mathnerd314>(g is using grep, which AFAIK only programmers can do)

<davexunit>as a developer, to update a package you do 2 things: change the version string, update the sha256 string of the source tarball.

<Mathnerd314>davexunit: you left out finding the version string, calculating the sha256, and testing the change (As I said, 14 steps)

<davexunit>Mathnerd314: I don't see your point. if it's that some of this process could be automated, then I agree.

<davexunit>but I don't think that upgrading a debian package is any easier.

<davexunit>and the point about users not knowing how to use grep is irrelevant, because we're talking about developers here.

<Steap>Mathnerd314: you seem to hate both Nix and Guix

<Mathnerd314>upgrading a debian package: apt-get install --only-updates <package>. And that took IIII=4 steps to find; if I was using Debian it'd probably be 1 or 2

<Mathnerd314>Steap: yes, I hate them because they are technically superior to Debian but, in practice, inferior.

<Steap>Mathnerd314: well, you sound a bit like "this shit sucks"

<Steap>so nothing is ever going to come out of this conversation

<davexunit>Mathnerd314: you just described the process of upgrading a package as user!

<davexunit>that is not what a developer does to update the package that is debian's repository.

<Steap>I suggest you propose some changes that would make Guix better

<Steap>or just leave :)

<Mathnerd314>davexunit: sure, but Debian has 3400 developers, so the cost of updating the package is spread out. In Guix there are 10 developers, so the cost is not spread out.

<Mathnerd314>davexunit: Debian's updating process could be 100x worse than Guix's and they would still have an advantage

<Mathnerd314>Steap: make updating easier than Debian by a factor of 1000. That's a change that would make Guix better than Debian.

<Steap>Mathnerd314: good, we'll carefully review your patch

<Mathnerd314>so that's where the issue comes in; upstream Nix is not patch-friendly

<Mathnerd314>indeed, so far as I can tell there have been no outside patches whatsoever (either written or accepted)

<davexunit>our lead maintainer has gotten patches into nix upstream.

<Mathnerd314>hmm, yes, let's look at them: https://github.com/NixOS/nix/commits?author=civodul

<davexunit>that's a good number of patches.

<Mathnerd314>I see changes to the builder, changes to nix-instantiate (which I remind you he later forked into Guix), and patches to build on GNU/Hurd

<Mathnerd314>there's nothing that actually modifies anything internal to Nix

<FracV>Mathnerd314: I'm not sure that's a good metric

<FracV>In many small projects the developers are people who are strongly and actively involved, and the projects have few passing patchers

<Mathnerd314>ok, so is there a better metric? "0" is indeed not very informative

<FracV>How about how well the software does the job you want it to do?

<FracV>Bonus points for being conceptually very clean

<Mathnerd314>but that is user-centric, which I was recently chastised for. as davexunit said, "we're talking about developers here"

<FracV>Then what do you *want*?

<FracV>Because you were complaining earlier that you haven't got anything development to do

<FracV>Because all the packages you want already exist

<FracV>I don't know why I'm even having this conversation. I'm barely involved with guix and just idle here for the news.

<FracV>You're just so frustratingly and pointlessly negative.

<Mathnerd314>indeed. As I said at the beginning, I had conflicting emotions. those often result in negativity. I was wondering if #guix had any advice

<FracV>It seems like you want an answer to an ill-defined problem

<Mathnerd314>not only that, an ill-posed problem

<Mathnerd314>although I think technically ill-defined and ill-posed are distinct

<Mathnerd314>as you can see from http://en.wikipedia.org/wiki/Well-posed_problem, a well-posed problem requires (1) a solution to exist (2) a unique solution and (3) continuous behavior

<Mathnerd314>an "ask for advice" problem satisfies none of these constraints

<Mathnerd314>on the other hand, despite this lack of well-posedness, many techniques have been developed for solving ill-posed problems. They are of course highly domain-specific.

<Mathnerd314>so asking in #guix was to see if anyone had developed these techniques and was willing to share

<Mathnerd314>FracV: I think you are mixing up me with my problem though; it is the *problem* that is frustrating and negative, not me

<Mathnerd314>I've been told on occasion that I'm a fun-loving very agreeable companion

<Mathnerd314>(although of course I could be biased)

<Mathnerd314>hmm, my numbers are off: http://www.perrier.eu.org/weblog/2014/07/27#devel-countries-201308 Debian has only 969 active developers, not 3400+

<FracV>A project being 1% of the size of Debian is probably quite good, then

<FracV>Windows has a lot of developers poached from top universities and it's not 1% of the quality of Debian

<Mathnerd314>Windows is an operating system; you are probably referring to Microsoft

<Mathnerd314>which according to http://www.microsoft.com/en-us/news/inside_ms.aspx has 64,641 "engineering-role employees"

<FracV>I was comparing OS to OS

<Mathnerd314>right, but to compare developers of OS's you have to actually count them

<Mathnerd314> http://www.forbes.com/sites/frederickallen/2012/07/03/the-terrible-management-technique-that-cost-microsoft-its-creativity/

<Mathnerd314>"out of every 10 people, 2 get a great review"

<Mathnerd314>and then they're cutting 14%: http://online.wsj.com/articles/microsoft-to-cut-18-000-jobs-1405599599

<Mathnerd314> http://www.zdnet.com/beyond-12500-former-nokia-employees-who-else-is-microsoft-laying-off-7000031726/

<Mathnerd314>"a number of Windows engineers, primarily dedicated testers, will no longer be needed."

<Mathnerd314>"The plan is for that team is to reduce our reliance on contingent staff augmentation by over 20 percent year-over-year"

<Mathnerd314>hmm. No clear numbers on how big the OS division is though.

<Mathnerd314>but I'm guessing their developers are somewhere in the 2000 range

<Mathnerd314>comparable to kernel + debian

<Mathnerd314>probably bigger by 1.2x or so

<Mathnerd314>anyways, the conclusion I draw is still that both Guix and NixOS are tiny

<Mathnerd314>and if you're tiny, you have to think like a startup: http://www.speakhuman.com/ch03_Startup_Thinking

<Mathnerd314>and fixing packages one-by-one is not the way to win

<Mathnerd314>(AFAICT)

<Steap>yeah, good, we're not a startup

<Steap>You do realise that you should be coding ?

<Mathnerd314>not really, I think time should be spent at least 50/50 on planning vs. coding

<Mathnerd314>why do you think you're not a startup?

<Steap>To be honest

<Steap>I have no idea of who you are

<Steap>you jsut came here and expect to tell us how to do stuff

<Steap>This is Free Software, you're welcome, but we care about how good you are, and should show that by writing actual code

<Steap>I think we're not a start-up because we're not a start-up

<Steap>that's that simple

<Steap>we're a bunch of devs hacking on a funny project

<Mathnerd314>I don't think I've told you how to do anything. I've asked for stuff to do and gotten "write packages". I've examined writing packages, and concluded it's pointless. So what's left?

<Steap>that's precisely my point

<Steap>I have no idea of who you are

<Mathnerd314>I suggested " make updating easier than Debian by a factor of 1000", and you said "we'll carefully review your patch". I interpreted that as sarcasm meaning that if I wrote a patch it would be rejected

<Mathnerd314>Steap: and I have no idea who you are, so we're even

<Steap>you came here to explain that Nix sucks and that, in the end, you don't like Guix and have no intention of doing anything

<Steap>well, I've been on this chan for quite a while

<sjamaan>Steap: Stop feeding the troll :)

<Steap>all I'm telling you is you either send us patches or you gtfo

<davexunit>hey everyone, upvote this on HN https://news.ycombinator.com/item?id=8096913

<Steap>sjamaan: not sure whether he's a troll since you've been feeding too :)

<davexunit>my post on release day didn't make it to the front page, but we're there now :)

<Steap>We should really get admin rights

<Steap>I think only civodul has them

<sjamaan>Steap: Yeah, but I stopped a few hours ago :)

<Mathnerd314>Steap: so are you saying that if I write a patch that does indeed making updating easier, it will be accepted?

<Steap>sjamaan: missed that

<Steap>Mathnerd314: we'll fucking read it, until then we're not feeding any more :)

<Mathnerd314>but I already submitted a patch (to a different project), and they read it and said "fuck your patch". The bug has been sitting in their bug tracker for 4 years now, open, with nobody fixing or reading it.

<Mathnerd314>"I'm not sure what the reasoning is behind this bug, really (or why the current code exists). What benefit is there to applying this patch?". Then there was a long discussion.. but no conclusion

<jxself>Submitting a patch to a project is no guarantee of acceptance. Actual acceptance depends on many factors.

<Mathnerd314>and AFAICT these factors are always "is submitter in my list of accepted submitters? if so, accept. otherwise, reject."

<Mathnerd314>which is not really helpful to new contributors

<Mathnerd314>(when they are trying to figure out what patch to write)

<FracV>Mathnerd314: Do you think it's possible you were unintentionally obnoxious?

<FracV>As much as FOSS people try not to let it factor in to their decision making, people skills are important

<sjamaan>The manual says you can unmount /mnt after installing, and reboot. However, when I try that, umount(8) says "device is busy"

<sjamaan>"deco stop cow-store" says "no service currently providing cow-store"

<FracV>sjamaan: Does lsof /mnt give any interesting output?

<sjamaan>It lists /mnt/tmp/guix-inst for unionfs

<jmd>The difference, is that with free software, if the maintains say "fuck your patch", you can always apply it yourself.

<FracV>sjamaan: I think that's you issue, though I don't know what to do about it

<sjamaan>I could kill it :)

<sjamaan>BTW, how does one safely reboot? There's no shutdown or reboot command available

<Mathnerd314>FracV: it's possible, but I don't know how to evaluate that (probably because of my lack of people skills)

<Mathnerd314>jmd: then I have to invest ongoing effort, which is nontrivial (my time is already spoken for by many others; even now, I should be working on other projects)

<Mathnerd314>I do have to invest ongoing effort anyways, but currently it has been relatively small (updating a few extensions)

<sjamaan>How was this image created? http://www.gnu.org/savannah-checkouts/gnu/guix/manual/images/bootstrap-graph.png

<sjamaan>Manually, or through some automated layout system?

<jmd>Mathnerd314: That is true. But you are able decide if you want to spend the effort or not.

<Mathnerd314>sjamaan: graphviz. nix-store --query --graph or something

<sjamaan>Thanks

<sjamaan>I should look into that

<sjamaan>still looking for a system to layout a complex database schema readably

<Mathnerd314>jmd: well, kind of. I can make the decision, but I can't determine if it was the correct one.

<jmd>Mathnerd314: Ahh no. If we could do that, then we would be gods.

<Mathnerd314>jmd: I've gone to a couple talks where people do modeling simulations of the world and figure out the right decisions. so it's definitely possible. But you're right, they used supercomputers which are unavailable to mere mortals :-)

<jmd>Supercomputers are used to predict the weather. Sometimes they are correct...

<Mathnerd314>yeah, there were weather talks, but also materials science and economics

<Mathnerd314>and biology, lots of biology

<Mathnerd314> http://www.siam.org/meetings/an14/ if you're wondering

<Mathnerd314>jmd: and all it required was using computers in the right way, not god-like qualitites

<Mathnerd314>and indeed, sometimes the models were wrong, but usually they caught that early and made them more complex

<jmd>... and then found the model was overfitting ...

<Mathnerd314>no, they used physics, which can't be overfitting, because it's been applied in too many domains

<jmd>Right. But there is a thing called the "no free lunch theorum"

<ijp>maybe, but we don't have any other universes to compare to

<jmd>How much disk space do I need to build gcc ?

<Mathnerd314>jmd: see http://en.wikipedia.org/wiki/No_free_lunch_in_search_and_optimization#Interpretations_of_NFL_results, 'The NFL results do not indicate that it is futile to take "pot shots" at problems with unspecialized algorithms.'

<Mathnerd314>they just show that it can't always work, in a mathematical sense

<jmd>t

<jmd>That is what I was trying to explain to you.

<Mathnerd314>jmd: http://www.linuxfromscratch.org/lfs/view/development/chapter06/gcc.html suggests it needs 2.5 GB

<Mathnerd314>jmd: but there's nothing in the NFL theorem that says you can't immediately determine your results are incorrect.

<jmd>I am getting this problem after building: cycle detected in the references of `/gnu/store/3gq9hcbdng1wp0zpmrxpqijv0djd12h0-gcj-4.8.3-lib'

<taylanub>is it fine to use Guile functions introduced later than 2.0.5 in package recipes? because the README etc. say version 2.0.5+ is needed for Guix, and I just noticed 2.0.5 not having `alist-delete' which comes in handy when inheriting from a package then removing some inputs

<Steap>taylanub: when was alist-delete added ?

<taylanub>I don't know... wonder if I did something else wrong?

<Steap>we use alist-delete all the time

<Steap>so I'd say it's fine

<ijp>taylanub: alist-delete is in srfi 1

<taylanub>oh, so I just add an import to the module in which I'll use it? (so far there's just a huge list of guix and gnu modules there so it seems a bit off)

<taylanub>never mind, I see other modules do it too

IRC channel logs

2014-07-28.log