samueldr changed the topic of #nixos-dev to: #nixos-dev NixOS Development (#nixos for questions) | NixOS 19.09 is released! | | | 19.09 RMs: disasm, sphalerite |
drakonis has quit [Quit: WeeChat 2.6]
<ryantm> Ericson2314:
<ryantm> Ericson2314: I'm not free at that time. I might be able to drop in for a few minutes if my kids sleep in.
<Ericson2314> ryantm: oh did I misread the thing?
<Ericson2314> I thought this was a time eveyrone was free
<ryantm> Ericson2314: I'm free at that time on Tuesday.
<Ericson2314> ryantm: can you go double check what you filled out?
<Ericson2314> so one thing that's confusing is the columns are not all the same day
<Ericson2314> if it hits midnight then it's the next day
<ryantm> oh
<Ericson2314> the headers are just for the top row until midnight
<ryantm> Sorry, I did it all wrong then.
* Ericson2314 wishes it didn't do that but just had more columns...
<Ericson2314> I did too :)
<ryantm> I'll fix it now
<Ericson2314> the black bars also confused me until I realized it was daylilght savings in the US
<clever> Ericson2314: i figured out the 9000 baud problem
<clever> Ericson2314: the newlib is compiled with the assumption that the cpu is running at 250mhz, but bootcode.bin is only running at 19.2mhz
<clever> Ericson2314: so everything (including the serial port) is running at ~1/13th speed
<Ericson2314> clever: hahahahha nice
<Ericson2314> new nix param { htz ? ... }
<Ericson2314> everyone will freak out
<clever> i'll just copy the baud init code into the main program
<clever> thats how avr typically handles it
<ryantm> Ericson2314: It would be nice if it told you the day/time when you hovered over a cell.
<Ericson2314> ryantm: it does on the aggregate side
<Ericson2314> along with who is available
<Ericson2314> but yeah it should on both sides
<ryantm> Ericson2314: I think I updated it correctly now.
<Ericson2314> ryantm: cool, thanks!
<ryantm> Unfortunately, it looks like there is no overlap.
<ryantm> I think it would be fine for you to keep your original meeting time if everyone else can meet.
<clever> Ericson2314: �Hello world!
<clever> Ericson2314: its alive!
<Ericson2314> clever: nice!
<Ericson2314> ryantm: I'll give people the option of that or something else in the email
<clever> [root@amd-nixos:~]# stty -F /dev/ttyUSB0 115200
<clever> Hello world!
<clever> Hello world!
<clever> [root@amd-nixos:~]# cat /dev/ttyUSB0
<clever> and it will repeat if i short out the RUN header (which is the master reset)
<{^_^}> #72657 (by cleverca22, 13 seconds ago, open): initial implementation of vc4 cross-compile
<Ericson2314> ryantm: heh it's taking a while because I'm finally getting around to migrating my calender + contacts to fastmail
<Ericson2314> getting invites on one mail and having to mvoe to a different calendar was quite annoying!
drakonis has joined #nixos-dev
<ryantm> Ericson2314: I just added like all of tomorrow as being available.
<Ericson2314> ryantm: OK cool
<ryantm> It looks like we a have huge block tomorrow as everyone available except gchristensen
<clever> Ericson2314: a helloworld example is also in the comments on the PR if you want to test
<Ericson2314> cool
<Ericson2314> ryantm: I will remind people of the misleading day columns in case that opens anything up
<clever> Ericson2314: some small design issues, my newlib was still using the custom gcc-stage1 and binutils, rather then the ones with src swapped
<clever> 11106 else if name == "newlib" && stdenv.targetPlatform.isVc4 then vc4-newlib
<clever> Ericson2314: how is this derivation supposed to get a copy of the stage1 cross-compiler?
<clever> gccCrossStageStatic in nativeBuildInputs fails
<Ericson2314> clever: check the other newlibs; there is stdenv for that
<Ericson2314> i forget what it is called, stdenvCrossStageStatic or something?
<clever> pkgs/development/misc/msp430/newlib.nix uses stdenvNoCC
<clever> and msp430GccSupport
<Ericson2314> clever: crossLibcStdenv;
<Ericson2314> grep for `crossLibcStdenv` and you will see
<Ericson2314> clever: someday we will convince gcc to let us build the libarries separately
<Ericson2314> or just do it anyways liek exherbo
<Ericson2314> and so much compleixty will go away
<Ericson2314> cc-wrapper will just let one choose their own libc, librt, libcxx, and we're good!
<clever> while evaluating the attribute 'stdenv' of the derivation 'newlib-vc4' at /home/clever/apps/nixpkgs-vc4/pkgs/development/misc/vc4/newlib.nix:4:3:
<clever> assertion failed at /home/clever/apps/nixpkgs-vc4/pkgs/top-level/all-packages.nix:7675:25
<clever> 7675 gccCrossStageStatic = assert stdenv.targetPlatform != stdenv.hostPlatform; let
<clever> 7668 crossLibcStdenv = overrideCC stdenv
<clever> 7669 (if stdenv.targetPlatform.useLLVM or false
<clever> 7671 else buildPackages.gccCrossStageStatic);
<clever> crossLibcStdenv just uses gccCrossStageStatic, so it doesnt solve the assertion i was having, it just makes a cc-wrapper for me
<Ericson2314> clever: oh see how the libcs are foo and fooCross
<Ericson2314> that might help
<Ericson2314> sorry this is the grossest part
<Ericson2314> I was able to get rid of all the *Cross derivations except for the libcs
<Ericson2314> it was very sad
<clever> 11106 else if name == "newlib" && stdenv.targetPlatform.isVc4 then targetPackages.vc4-newlib
<clever> Ericson2314: this seems to have helped
<Ericson2314> clever: don't forget the or xxxx
<Ericson2314> cargo cult to the max!
<clever> wasnt obvious why thats being done
<Ericson2314> clever: it is so you can refer to to them in the stage with all the other libraires
<Ericson2314> because targetPackages.targetPackages is a runt stage, whereas it probably shouldn't be
<Ericson2314> more grossness!
<clever> copying path '/nix/store/zc1zzcl9cb2wx220v7d2m4pxxas1rddl-gcc-debug-8.3.0-vc4-elf-stage-static' from 'ssh://builder@'...
<clever> checking host system type... Invalid configuration `vc4-elf': machine `vc4-unknown' not recognized
<clever> hmmmm, now newlib is busted
<clever> builder for '/nix/store/38b7njh6zflh96s29q0cvhlmn7g1vz41-newlib-vc4-vc4-elf.drv' failed with exit code 1
<clever> configure flags: --disable-static --prefix=/nix/store/945d9lb72kymsig94s85v53wgm4w814v-newlib-vc4-elf --target=vc4-elf --build=x86_64-unknown-linux-gnu --host=vc4-elf
<clever> 12 configureFlags = [ "--target=vc4-elf" ];
<clever> Ericson2314: i dont think this newlib supports --host=vc4-elf, how would i tell the stdenv to not do that?
<Ericson2314> clever: what is the error?
<clever> configure flags: --disable-static --prefix=/nix/store/945d9lb72kymsig94s85v53wgm4w814v-newlib-vc4-elf --target=vc4-elf --build=x86_64-unknown-linux-gnu --host=vc4-elf
<Ericson2314> see what the other newlibs do
<clever> checking build system type... x86_64-unknown-linux-gnu
<clever> checking host system type... Invalid configuration `vc4-elf': machine `vc4-unknown' not recognized
<clever> configure: error: /nix/store/v2x1ks46fq3yn8c94qfv38wrvmr5xn5r-bash-4.4-p23/bin/bash ./config.sub vc4-elf failed
<clever> builder for '/nix/store/vx862y2hylhzi6lr6s0imif81vkibiyj-newlib-vc4-elf.drv' failed with exit code 1
<clever> 20 configureFlags = [
<clever> 21 "--host=${stdenv.buildPlatform.config}"
<Ericson2314> clever: pkgs/development/misc/newlib/default.nix that configure platform line
<clever> yeah, thats where i got the last 2
<clever> it worked before, when --host wasnt being passed
<Ericson2314> clever: seems related too
<Ericson2314> but with `configurePlatforms = [ "build" "target" ];` it won't pass --hsot
<Ericson2314> *--host
<clever> ah, was just trying to reverse engineer that var
<clever> still failing though, *investigates more*
octe has quit [Ping timeout: 246 seconds]
<clever> hmmm, it has a vc4 binutils and gcc in PATH
nbp has quit [Ping timeout: 244 seconds]
<Ericson2314> that's good, right?
<clever> yeah
<clever> checking target system type... Invalid configuration `vc4-elf': machine `vc4-unknown' not recognized
<clever> but this error makes even less sense, everything should be working
nbp has joined #nixos-dev
<clever> when things worked, it was doing:
<clever> configure flags: --disable-static --prefix=/nix/store/hsy72n0caycw2w8ycvlznb137cfxbk63-newlib-vc4 --target=vc4-elf
<clever> now when failing, its doing:
<clever> configure flags: --disable-static --prefix=/nix/store/i8qsn0nb6rn6hgdf3k5nbyxgjzj8qcn6-newlib-vc4-elf --target=vc4-elf
<clever> working, checking target system type... vc4-unknown-elf
<clever> broken, checking target system type... Invalid configuration `vc4-elf': machine `vc4-unknown' not recognized
<Ericson2314> clever: is the update gnu config thing going?
<clever> ?
<Ericson2314> there is a setup hook update config.sub
<Ericson2314> uptream gnu config doesn't have vc4
<Ericson2314> maybe the patched source has it, but the setup hook wiped it
<clever> Updating Autotools / GNU config script to a newer upstream version: ./config.sub
<Ericson2314> yeah
<clever> yeah, this msg wasnt in the working build
<Ericson2314> i would send a patch upstream
<clever> dontUpdateAutotoolsGnuConfigScripts for testing
<Ericson2314> whether or not gcc supports it on mater, gnu config purports to know about everything
<Ericson2314> sure
<clever> checking target system type... vc4-unknown-elf
<clever> got past that
<clever> configure: error: C compiler cannot create executables
<Ericson2314> weird, you are building libc so it shouldn't expect that
<clever> 112 /nix/store/rr7860kh97077r72mvrdlp8yp5yj7vnr-vc4-elf-binutils-2.31.1/bin/vc4-elf-ld: cannot find -lc
<clever> oh right
<Ericson2314> hehehe
<clever> 14 # newlib expects CC to build for build platform, not host platform
<clever> it wants a an x86 gcc, to make build tools
<Ericson2314> clever: any idea how old the fork is?
<Ericson2314> that's unfortunately, that should be called CC_FOR_BUILD
<clever> the fork itself, was last touched 3 years ago
<clever> and the newest commit from somebody other then the forker, is from march of 2015
<clever> 87 ./configure: eval: line 1: cc: not found
<clever> Ericson2314: crossLibcStdenv lacks a cc for x86!
<clever> depsBuildBuild = [ ]; i'm guessing
<Ericson2314> yeah
<clever> the --host likely wasnt to blame
<clever> it was config.sub
<Ericson2314> cool
<Ericson2314> that's a good sign
<clever> and i somehow deleted nativeBuildInputs by accident, so textinfo vanished
<clever> building '/nix/store/304bk6zfhvi7zk2c95a120bqqpwyvrp5-gcc-debug-8.3.0-vc4-elf-stage-final.drv'...
<clever> Ericson2314: newlib builds once more!, on to stage2 gcc
<Ericson2314> clever: nice!
<Ericson2314> and hopefully that works fine since it did before
<clever> i also threw together a drv to run it under resim
octe has joined #nixos-dev
<clever> [clever@amd-nixos:~/apps/rpi/hello-world]$ nix-build -I nixpkgs=../../nixpkgs-vc4/ --show-trace -A testit --option builders '' --keep-failed
<clever> building '/nix/store/4wd7hw5xi2sizxgjwkqk4ws775k2zlhc-testit.drv'...
<clever> Hello world!
<clever> /nix/store/d2cvzfl5jvizwgf4qdlljjqiwyq9lz8g-testit
<clever> Ericson2314: and it works!
<Ericson2314> clever: great!
<clever> Ericson2314: pushed to my PR
* clever heads off to bed
<Ericson2314> sweet
drakonis has quit [Quit: WeeChat 2.6]
ris has quit [Ping timeout: 258 seconds]
<MichaelRaskin> Ericson2314: I think there is also an option of posting a meeting link in the PR discussion.
<gchristensen> ryantm, Ericson2314 unfortunately I'm not likely to be available 6pm to midnight here today, as it is my only weekend in town and I'm trying to make use of it
<MichaelRaskin> It looks like there is no time when I am the only or one of the two last people to be a problem, so I guess I can just wait and see
<gchristensen> is there no overlap?
<MichaelRaskin> Apparently not
<gchristensen> ouch
<MichaelRaskin> There are some 4/5
<MichaelRaskin> As a nonbeliever in voice chats I can start wondering whether a commitment to answer mentions allows us to run the same discussion as a thread spending a week wall-clock time (which we are apparently going to spend anyway just on finding a time slot) and probably _less_ active time per person
<gchristensen> how about 10pm today
<MichaelRaskin> You mean 22:00 UTC?
<gchristensen> yeah
<MichaelRaskin> Seems fine in the table (and fine with me)
phreedom_ has joined #nixos-dev
phreedom has quit [Ping timeout: 260 seconds]
FRidh has joined #nixos-dev
drakonis has joined #nixos-dev
Scriptkiddi1 has joined #nixos-dev
Willi_Butz has joined #nixos-dev
orivej has joined #nixos-dev
MichaelRaskin has quit [Ping timeout: 265 seconds]
WilliButz has quit [Ping timeout: 265 seconds]
Scriptkiddi has quit [Ping timeout: 265 seconds]
__monty__ has joined #nixos-dev
srhb_ has joined #nixos-dev
srhb_ is now known as srhb
orivej has quit [Ping timeout: 240 seconds]
ris has joined #nixos-dev
Willi_Butz is now known as WilliButz
orivej has joined #nixos-dev
MichaelRaskin has joined #nixos-dev
<worldofpeace> Hey, so I was told I'd RM 20.03, but no one's gotten back to me on this.... sphalerite disasm
<qyliss> worldofpeace++
<{^_^}> worldofpeace's karma got increased to 42
<ryantm> worldofpeace++
<{^_^}> worldofpeace's karma got increased to 43
<ryantm> 22:00 UTC works great for me.
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos-dev
orivej has quit [Ping timeout: 240 seconds]
drakonis_ has joined #nixos-dev
tilpner has joined #nixos-dev
drakonis has quit [Ping timeout: 240 seconds]
drakonis_ has quit [Ping timeout: 250 seconds]
drakonis_ has joined #nixos-dev
srhb has quit [Changing host]
srhb has joined #nixos-dev
asymmetric has quit [Changing host]
asymmetric has joined #nixos-dev
drakonis has joined #nixos-dev
drakonis_ has quit [Ping timeout: 265 seconds]
drakonis1 has joined #nixos-dev
<qyliss> Where do these updateScripts written in Nix come from?
<qyliss> I find them extremely difficult to read, and it feels like a bit of a
<qyliss> hack to be writing actual programs like this in Nix
<worldofpeace> There's updateScripts in nix?
<qyliss> nixpkgs
<qyliss> sorry
<worldofpeace> I don't think any of those updateScripts in passthru could ever be implemented in nix, I believe it needs some sort of script to run.
<qyliss> Look at Chromium's update.nix
<worldofpeace> right it does nix-build --builders "" -Q --no-out-link update.nix -A update
<qyliss> Yeah
<qyliss> But the whole of the functionality is implemented in Nix
<qyliss> With loads of IFD and stuff
<worldofpeace> Right, here's the commit that did the rewrite And right, IFD and nix, doesn't this make it a pretty slow update?
<qyliss> For that particular one, I suspect the slow part will be Google's server
<qyliss> Having just written an update script for crosvm in Python that uses the same server and takes several seconds to run
<worldofpeace> not that I'd change it, it's just an update script. I typically try to keep them in bash because they don't do much tbh.
<qyliss> Some of them do, like my crosvm one
<qyliss> Because Google is so bad at versions
<qyliss> The Chromium one is also extremely complicated AFAICT
<qyliss> although I have not yet managed to figure out entirely what it's actually doing.
<qyliss> I wouldn't like to write that in bash, but it would be better than Nix...
<qyliss> > Along the way, this has led to a quite massive workaround, which abuses
<{^_^}> error: syntax error, unexpected ',', expecting ')', at (string):269:14
<qyliss> MD5 collisions to detect whether an URL is existing
<qyliss> WHAT
<qyliss> :o
<worldofpeace> Yeah, my eyes were suddenly fatiqued once trying to parse it. If we had to sort langs for update scripts I'd say python wins. case in point is implemented in :D
<MichaelRaskin> Ah, qyliss discovers the joys of MD5 abuse
<MichaelRaskin> I think we should upgrade it to the known SHA1 collision now!
<qyliss> My use case here is that I'd like to introduce a new chromium channel, for puppeteer
<samueldr> qyliss: I think that specific script is more about showing off than actual good implementation :)
<qyliss> There's demand for it:
<{^_^}> #71179 (by robinp, 2 weeks ago, open): chromium versions that match `puppeteer` release versions.
<qyliss> But I'm wondering if rather than trying to add it to that script, I should just rewrite the script from scratch.
<qyliss> Because as cool as that script is, it's not maintainable
<qyliss> I think the answer I'm getting might be yes.
<samueldr> I'm not saying no :)
<worldofpeace> qyliss: You've read my mind, it's one I wouldn't want to touch
<samueldr> sounds like making something more maintainable is a good thing
<qyliss> Well, sounds like I at least have somebody to approve my PR :)
<MichaelRaskin> Ah right.
<MichaelRaskin> Of course on that day
<worldofpeace> lol, in fact it's aesthetically beautiful qyliss acording to so it sounds like there's an attachment...
<qyliss> I'm a big fan of it as an art piece
<qyliss> And it will be forever preserved in the git log
<MichaelRaskin> You just don't want this code to be executed outside a sandbox?
<infinisil> qyliss: Check out `git log --reverse ea2fbd786415..8f29e4edb3ec`
<infinisil> (if you haven't seen it already)
<gchristensen> MichaelRaskin, Ericson2314 are we actually meeting today? I'm very sleepy :/
<qyliss> infinisil: what am I looking at in that?
<worldofpeace> hmm, perhaps they're a dilettante
<qyliss> Oh right
<Ericson2314> gchristensen: we can do another time
<worldofpeace> infinisil: that was actually pretty enjoyable to look at :D
<MichaelRaskin> gchristensen: I am more or less completely lost by now
<Ericson2314> if you are sleepy + missing person, that setttles it for me
<qyliss> aszlig++
<{^_^}> aszlig's karma got increased to -666
<qyliss> lmao
<FRidh> haha
<infinisil> xD
<Ericson2314> MichaelRaskin: gchristensen I'll make one with more days
<MichaelRaskin> Ericson2314: Write an email to announce failure first
<MichaelRaskin> (my claim from today's morning has just got confirmed…)
<Ericson2314> ok
<Ericson2314> MichaelRaskin: I don't want to just decree we are meeting without someone too bad, or we are meeting on crazy short notice; sorry if my tip-toing just added more confusion
<MichaelRaskin> No, another claim
<MichaelRaskin> I said that an email thread with a commitment to answer all direct mentions from each other might end up being faster.
<worldofpeace> infinisil: your bot is going to frighten people with these values
<gchristensen> worldofpeace: it is aszlig's doing :)
<samueldr> worldofpeace: wanna see something spooky? aszlig++
<{^_^}> aszlig's karma got increased to -666
<worldofpeace> gchristensen: I appreciate this humor then 🤣
<samueldr> (there's automation on aszlig's part to reduce the karma)
<aszlig> samueldr: unfortunately it's not X-D
<samueldr> oh, you sneaky
<gchristensen> just deep commitment
<aszlig> samueldr: was too lazy to automate it, but if it gets too anoying i might do it :-D
<aszlig> qyliss: there are other ascii thingies in the git log btw :-D
<worldofpeace> Queue aszlig appreciation thread 😂 just you know, ignore any comments about chromium's update script
<aszlig> worldofpeace: well, ideally the updater would be written in something else than nix
<Ericson2314> aszlig: voth said you were the person to look at my crazy windows cross stuff?
<aszlig> Ericson2314: uh oh... windows X-D
<aszlig> Ericson2314: i actually saw that issue in my mails but didn't look at it yet, because it said "windows" =)
<Ericson2314> very fair :)
<aszlig> Ericson2314: and it has been years since i did anything with windows cross-builds and runInWindowsVM
<worldofpeace> aszlig: was nix the only better option at the time...
<Ericson2314> ah ok
<samueldr> I bet on s/better/fun/
<Profpatsch> aszlig: oh man, that’s haskell packaging in 2012, we’ve come a long way :D
<aszlig> worldofpeace: is that a question? :-D
<worldofpeace> Lol, everything is full 🔵 now
<MichaelRaskin> When2Meet table looks a bit abstract-artsy now…
* Ericson2314 takes a look
<Ericson2314> MichaelRaskin: ah i forgot sunday the 9th
<Ericson2314> must preserve translational symmetry!
<gchristensen> dang. nix doesn't work if HOME=/homeless-shelter and you import a tar.gz URL
<qyliss> Oh yeah I had that problem with my ro home
<infinisil> Am somewhat jelly of your ro home
<gchristensen> same
<gchristensen> qyliss: is the fix adding rw location?
<qyliss> It needs XDG_CACHE_HOME
<qyliss> specifically $XDG_CACHE_HOME/nix needs to be read-only
<qyliss> read-write, rather
<infinisil> Ah nice, probably all of XDG_CACHE_HOME would make sense to be rw
<infinisil> And could even wipe it automatically
<qyliss> I map XDG_CACHE_HOME and XDG_DATA_HOME to ~/state
<qyliss> ~/state is ro because I want to review what's writing there
<qyliss> And make sure it's behaving well and not writing outside of its own subdir
<gchristensen> ah nice
<gchristensen> thanks, qyliss
<qyliss> gchristensen: may I PM, btw?
<gchristensen> yep
FRidh has quit [Quit: Konversation terminated!]
<gchristensen> I'm thinking it'd be rude if ofborg startedbuilding Nix prs outside of the RFC workflow?
<infinisil> outside of the rfc workflow?
<gchristensen> yeah, there is an RFC for building Nix PRs
<infinisil> Ah you mean to just make it build things already even if the rfc isn't accepted (yet)?
<gchristensen> right, or in ways the RFC doesn't specify
<infinisil> It's probably fine, I can't think of any reason anybody would dislike that
<Ericson2314> qyliss: have you heard of people being unable to join #spectrum with matrix?
<tilpner> gchristensen: Perhaps don't build those marked as drafts/WIP
<gchristensen> yeah
<qyliss> q3k: no??
<gchristensen> tilpner: I think ofborg already does this
<tilpner> gchristensen: But otherwise I don't consider it rude at all
<qyliss> err, Ericson2314
<qyliss> Not sure how I typoed that
<qyliss> Sorry q3k
<qyliss> Ericson2314: what goes wrong?
<Ericson2314> it says some 502 failure, not a very informative error message
<qyliss> Ericson2314: I've been told that the people to talk to about that are the Matrix IRC appservice people
<qyliss> I don't know anything about Matrix, I'm afraid
<Ericson2314> @qyliss thanks though for looking into it
<Ericson2314> I'll ask them
* gchristensen thinks the necessary changes to ofborg are nearly complete
rycee has quit [*.net *.split]
Taneb has quit [*.net *.split]
fadenb has quit [*.net *.split]
rycee has joined #nixos-dev
Taneb has joined #nixos-dev
fadenb has joined #nixos-dev
LnL has joined #nixos-dev
LnL has quit [Changing host]
LnL has joined #nixos-dev
tilpner has quit [Quit: tilpner]
orivej has joined #nixos-dev
drakonis1 has quit [Quit: WeeChat 2.6]
drakonis1 has joined #nixos-dev
cptchaos83 has quit [Quit: - Chat comfortably. Anywhere.]
cptchaos83 has joined #nixos-dev
__monty__ has quit [Quit: leaving]
orivej has quit [Ping timeout: 240 seconds]