worldofpeace changed the topic of #nixos-dev to: NixOS Development (#nixos for questions) | NixOS 20.09 Nightingale ✨ https://discourse.nixos.org/t/nixos-20-09-release/9668 | https://hydra.nixos.org/jobset/nixos/trunk-combined https://channels.nix.gsc.io/graph.html | https://r13y.com | 20.09 RMs: worldofpeace, jonringer | https://logs.nix.samueldr.com/nixos-dev
<lukegb> another option rather than using tsvector is to use a trigram index
<lukegb> then you can use the includes-regex thing
<aszlig> primeos: hm, you seem to be doing most of the chromium-related work, are you okay, if i post your contact info on https://chromium.googlesource.com/chromium/src/+/refs/heads/master/docs/linux/chromium_packages.md
<aszlig> ?
<aszlig> or do we have some other address that is non maintainer-related which we can use instead?
<aszlig> gchristensen: btw. i wonder when pkgs.chromium stopped working, since i explicitly used pkgs. to avoid the nixos tests to show up in the search query
<aszlig> probably the jobset is hidden nowadays
<samueldr> nixpkgs vs. nixos jobsets?
<aszlig> samueldr: the search is global
<aszlig> so it shouldn't matter
<samueldr> yeah, but jobs have different names
<aszlig> yeah
<aszlig> well, i don't remember anymore since it has been a while
<samueldr> nixos jobset: nixos.tests.chromium.x86_64-linux nixpkgs.chromium.x86_64-linux and nixpkgs jobset: chromium.x86_64-linux
<samueldr> though possibly things have changed since... that while
<aszlig> i think nixpkgs.chromium.x86_64-linux sounds about right
bennofs_ has joined #nixos-dev
bennofs__ has quit [Ping timeout: 240 seconds]
<gchristensen> we've grown by about 100,000,000 builds since then
<aszlig> well, anyway, it would at least be a good idea if people who *actually* maintain and probably also use chromium on nixos would be the contact person :-D
stolyaroleh_ has quit [Ping timeout: 265 seconds]
rajivr has joined #nixos-dev
supersandro2000 has quit [Disconnected by services]
supersandro2000 has joined #nixos-dev
orivej has quit [Ping timeout: 265 seconds]
mkaito has quit [Quit: WeeChat 3.0]
cole-h has joined #nixos-dev
bgamari has quit [Ping timeout: 256 seconds]
bgamari has joined #nixos-dev
bgamari_ has joined #nixos-dev
bgamari has quit [Ping timeout: 264 seconds]
kalbasit_ has joined #nixos-dev
<gchristensen> samueldr: I can return a search producing no results in 200ms now
<samueldr> great!
<samueldr> though at github's scale it should be a unicorn in 30 seconds flat
<samueldr> wait
<samueldr> is that 200ms to *confirm* there are no results?
<samueldr> like for search queries that ought to return nothing?
<samueldr> or is it returns nothing in 200ms while it should have?
<samueldr> the former being way more exciting
<gchristensen> I'm pretty sure it is confirming there are no results
<gchristensen> yp
kalbasit_ has quit [Ping timeout: 240 seconds]
<samueldr> oh, neat
<gchristensen> the overall page loads in 7s
<gchristensen> not better than what is there already in the success case
<gchristensen> I wonder why not
<lukegb> gchristensen: what other indexes are on BuildOutputs.path?
<siraben> Is this `python = python3` redundant? `mod_wsgi3 = callPackage ../servers/http/apache-modules/mod_wsgi { python = python3; };`
<lukegb> e.g. what does the perf of `WHERE path ~ '.*(pkgs\.chromium).*'` look like if you have a `CREATE INDEX expIndexBuildOutputsPathGinTrgm ON BuildOutputs USING GIST (path gist_trgm_ops);`
<lukegb> (although you might need to load pg_trgm first)
<gchristensen> lukegb: is it possible to have a GIN index where an integer field is also part of the index?
<lukegb> I'm not sure, I don't *think* so
<gchristensen> me either
<gchristensen> it is using the pkey instead of the gin index, I think because of a join
<gchristensen> it is weird, lukegb, explain analyze SELECT * FROM buildoutputs WHERE to_tsvector('english', regexp_replace(buildoutputs.path, '[\./-]', ' ', 'g')) @@ websearch_to_tsquery('english', regexp_replace('chromium', '[\./-]', ' ', 'g')) limit 10; -> seq scan on buildoutputs, if I drop the limit it uses the gin
<gchristensen> oh the table is too large to have 1 bit per row and fit in work_mem making it a lossy bitmap index
<gchristensen> that certainly makes it interesting :)
<gchristensen> something to play with in the morning :)
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-dev
orivej has joined #nixos-dev
orivej has quit [Ping timeout: 264 seconds]
orivej has joined #nixos-dev
cole-h has quit [Ping timeout: 264 seconds]
<samueldr> I can't wrap my head around that
tilpner has quit [Remote host closed the connection]
tilpner has joined #nixos-dev
<siraben> samueldr: what's going on there?
orivej has quit [Read error: Connection reset by peer]
<samueldr> exactly
orivej has joined #nixos-dev
Mic92 has joined #nixos-dev
Mic92 has quit [Quit: WeeChat 3.0]
Mic92 has joined #nixos-dev
krkini has joined #nixos-dev
AlwaysLivid has quit [Remote host closed the connection]
AlwaysLivid has joined #nixos-dev
<domenkozar[m]> ugh, my nix-daemon was using 31G of memory
LnL has quit [Read error: Connection reset by peer]
<domenkozar[m]> ah, dumping very large path
<domenkozar[m]> no way of knowing what
justanotheruser has quit [Ping timeout: 272 seconds]
<primeos> aszlig: yeah, feel free to post my contact info there - and thanks for your great and very impressive work on Chromium
<primeos> (and the hash collisions for the update.nix script where quite fun/legendary xD)
endocrimes has joined #nixos-dev
<domenkozar[m]> if only we had a nix errors project :)
<siraben> domenkozar: what would a nix errors project resolve?
endocrimes has quit [Quit: running from the computers]
<siraben> supersandro2000: could you explain why removing `python = python3` treewide is good? Is it because `python` defaults to it anyway?
endocrimes has joined #nixos-dev
<andi-> I wish we had learned that just python without a version will bite us on the next migration.
<V> Why did we have python = python3 in the first place? The huge 2->3 mess should have taught us something
<V> Wondering how many distros made /usr/bin/python a symlink to /usr/bin/python3 now, vs just having it remain a symlink to python2
<V> siraben: did you see my other comments re poetry btw
<yorick> V: we had python = python2 in the first place
<siraben> V: on the PR in #nixos?
<siraben> s/in/or
<V> siraben: yes
<V> yorick: "in the first place" = at any point, not what it pointed to initially
<siraben> V: just saw it now
<siraben> ok but I think rmrl is both a library and an application, hm
<V> Or, wait, it doesn't point to python3?
<siraben> V: it would be used like `callPackage ../foo { python = python3; }`
<siraben> it should be converted to python3 in ../foo/default.nix as well right?
<V> nvm me, I thought you were talking about something else in that case
<yorick> V: afaict it still points to python2
<V> yorick: yes, I misunderstood what siraben was saying
<yorick> it made sense to do back in 2014
<V> I thought this was about python at the top level, rather than as an override
<yorick> I hope python has learned from their 2->3 fiasco and python 4 will be backward compatible, and in that case it would be fine to python = python3
<yorick> I'd rather see no python alias, but if we must, then python = python3
<V> siraben: doesn't look like it's anything but a library
<yorick> iirc roommate was working on this a while ago
<siraben> xdg_utils → xdg-utils naming needs to be done treewide it seems
<yorick> those modules with the missing types sure weren't the most beautiful modules
<yorick> I wonder if we should just fix the missing types and then move on until someone else discovers them again
orivej has quit [Ping timeout: 246 seconds]
orivej has joined #nixos-dev
<yorick> whaaat, no stdenv.lib anymore?!
<yorick> ok
orivej has quit [Ping timeout: 264 seconds]
AlwaysLivid has quit [Ping timeout: 268 seconds]
<infinisil> > stdenv.lib
<{^_^}> { __unfix__ = <LAMBDA>; add = <CODE>; addContextFrom = <CODE>; addErrorContext = <CODE>; addErrorContextToAttrs = <CODE>; addMetaAttrs = <CODE>; all = <PRIMOP>; and = <CODE>; any = <PRIMOP>; appendToN...
<infinisil> (Doesn't show traces)
<jtojnar> > builtins.trace "foo" null
<{^_^}> null
<jtojnar> > throw "foo"
<{^_^}> foo
<jtojnar> can we disable aliases on the bot? then this should do https://github.com/NixOS/nixpkgs/pull/111447
<{^_^}> #111447 (by siraben, 48 minutes ago, open): stdenv/generic: throw when using stdenv.lib and disallowing aliases
<siraben> yorick: `stdenv.lib` was a blight upon nixpkgs
<siraben> granted it wasn't maliciously added or anything 13 years ago
<siraben> it was for convenience
__monty__ has joined #nixos-dev
stolyaroleh_ has joined #nixos-dev
Mic92 has quit [Quit: WeeChat 3.0]
Mic92 has joined #nixos-dev
Mic92 has quit [Client Quit]
Mic92 has joined #nixos-dev
<supersandro2000> siraben: python defaults to python2 IIRC and it removes complexity that is not required.
s1341 has joined #nixos-dev
<s1341> hi
<s1341> i'm trying to build a cross-toolchain for aarch64 android.
<s1341> in aarch64-unknown-linux-gnu-stage-final-gcc-debug-wrapper-6.5.0, libgcc dies because it tries to use /lib/cpp as the c-preprocessor.
<s1341> any ideas as to how I should fix this?
<supersandro2000> nice
<{^_^}> #111431 (by andir, 2 hours ago, open): [staging] autoconf: 2.70 -> 2.71
<supersandro2000> stdenv.lib got thrown in ofborg!
<supersandro2000> not nice. It is totally unrelated
dhess has quit [Quit: ZNC - https://znc.in]
<s1341> supersandro2000: any ideas about my /lib/cpp issue?
<sterni> supersandro2000: should probably revert the change to make ofborg useful again :|
<supersandro2000> s1341: nope
<supersandro2000> sterni: it is one usage that somehow sneaked in
<supersandro2000> should be fixed with https://github.com/NixOS/nixpkgs/pull/111472
<{^_^}> #111472 (by SuperSandro2000, 58 seconds ago, open): audit: Remove usage of stdenv.lib
<gchristensen> how did it sneak in? can you find the commit that introduced it?
<s1341> anyone?
<gchristensen> supersandro2000: ^
<supersandro2000> probably stale ofborg
<gchristensen> indeed, cool
<s1341> anyone know who deals with cross-compilation issues?
<supersandro2000> Mic92 maybe can help you
mkaito has joined #nixos-dev
mkaito has joined #nixos-dev
<s1341> @Mic92 you around?
<s1341> thanks @supersandro2000
<supersandro2000> I pinged him already. No need to ping him twice
<s1341> thanks supersandro2000
<stolyaroleh_> Hi @supersandro2000, do you know who is a good person to speak about fixing lldClang? It uses a flag that ld-wrapper.sh misses, resulting in binaries that can't run if you just try to use it in your stdenv.
<stolyaroleh_> I have two ideas on how to fix, not sure which one is better.
krkini has quit [Remote host closed the connection]
<gchristensen> maybe LnL
kini has joined #nixos-dev
<Mic92> s1341: No doing sports
<s1341> Mic92: ?
<siraben> supersandro2000: yay
<siraben> supersandro2000: what's stale ofborg?
bennofs has joined #nixos-dev
bennofs has quit [Client Quit]
bennofs has joined #nixos-dev
<supersandro2000> siraben: it succeeded a while ago (like 3 months) but wouldn't now
<siraben> ah
<supersandro2000> Does someone know if 20.09 is currently blocked?
<supersandro2000> When I build a container with it I get https://github.com/NixOS/nixpkgs/issues/107216
<{^_^}> #107216 (by worldofpeace, 6 weeks ago, closed): documentation.info.enable broken
<supersandro2000> building busybox and putting it into sandbox-paths fixes it
<supersandro2000> but I need to edit the read only file outside of the container
<supersandro2000> well didn't work...
<supersandro2000> needed to set it on the host nix-daemon...
cole-h has joined #nixos-dev
orivej has joined #nixos-dev
<supersandro2000> I hope staging is fixed now
<supersandro2000> there are some packages which do `doBuild = false;``. Does that option exist or should it be converted to ``dontBuild = true;``?
<gchristensen> probably shouldn't convert them, but just delete the definition
cole-h has quit [Ping timeout: 264 seconds]
saschagrunert has joined #nixos-dev
Raito_Bezarius has joined #nixos-dev
lassulus_ has joined #nixos-dev
kalbasit_ has joined #nixos-dev
lassulus_ is now known as lassulus
ScottHDev5 has joined #nixos-dev
<ScottHDev5> Hello, I don't understand why I don't have the --expr flag on nix eval documented here https://nixos.wiki/wiki/Nix_command/eval
<ScottHDev5> I get error: unrecognised flag '--expr
kalbasit_ has quit [Ping timeout: 260 seconds]
saschagrunert has quit [Remote host closed the connection]
rajivr has quit [Quit: Connection closed for inactivity]
<infinisil> ScottHDev5: It only exists in newer versions
<infinisil> Use --help to see what flags your current version supports
lassulus_ has joined #nixos-dev
lassulus has quit [Ping timeout: 240 seconds]
lassulus_ is now known as lassulus
<supersandro2000> what is https://search.nix.gsc.io/? A fancy software for searching?
<supersandro2000> is it better then sourcegraph?
<supersandro2000> going to remove it then
<adisbladis> supersandro2000: Thats hound
<supersandro2000> that hound http://www.houndsoftware.net/?
<supersandro2000> the name is not ideal tbh
itai has joined #nixos-dev
<itai> hey, is nix 2.4pre stable enough for end-users? or should i hold off unill it gets released?
<itai> wait shit i should ask this in #nixos shouldn't i, my bad
<adisbladis> > hound.meta.homepage
<adisbladis> supersandro2000: ^
<supersandro2000> itai: it should work most of the time
<supersandro2000> adisbladis: thanks
<lukegb> it's probably better than sourcegraph because sourcegraph seems to be bad at indexing top-level/all-packages.nix
<supersandro2000> itai: the commands are a bit different and darwin support sometimes breaks but the normal command work like a charm
<supersandro2000> Can someone with team creating experience take a look at https://github.com/NixOS/nixpkgs/pull/110623 ?
<{^_^}> #110623 (by yurrriq, 1 week ago, open): maintainers/teams: add beam team
<itai> supersandro2000: yeah the main reason i want to use it is because of the experimental commands, and flake stuff
<itai> they seem to be the way of the future so i figured i might as well learn them
<itai> does nix 2.4 have a target release date? or is it just a when it's ready type thing. also are questions welcome here or not? i've gotten mixed signals about that
<niksnut> hopefully within a few weeks
<niksnut> it's mostly a matter of writing release notes
<itai> neat
<niksnut> though there is at least one crash bug that needs to be sorted out
<adisbladis> itai: This channel is not so much intended for end-user questions but for nix/nixpkgs hacking. Generally default to #nixos.
<itai> adisbladis: right, i'll go there in the future, sorry for the fuss
<adisbladis> But don't hesitate to ask questions, worst case someone will guide you to the correct channel
<itai> yeah for sure
<adisbladis> Nothing to be sorry about :)
itai has quit [Quit: #nixos]
v0|d has joined #nixos-dev
mkaito has quit [Quit: WeeChat 3.0]
orivej has quit [Ping timeout: 240 seconds]
orivej has joined #nixos-dev
<andi-> gchristensen: are you working on hydra again? It times out here
kalbasit_ has joined #nixos-dev
kalbasit_ has quit [Ping timeout: 256 seconds]
<ScottHDev5> infinisil Ok thanks
<supersandro2000> staging evals now again
<gchristensen> andi- no
kalbasit_ has joined #nixos-dev
<andi-> gchristensen: mhm, can you confirm that the webinterface is unreachable?
<andi-> ahh it is back now
kalbasit_ has quit [Ping timeout: 246 seconds]
<supersandro2000> was down or really slow for me, too
<supersandro2000> or better is. Still not loading for me.
<gchristensen> I'm looking in to some things to do
<{^_^}> firing: RootPartitionLowDiskSpace: https://monitoring.nixos.org/prometheus/alerts
<gchristensen> I cut back the scrape time for prometheus, which was hitting it pretty often
<gchristensen> I'm thinking about writing an apache rule which cancels GET /search?query=pkgs.chromium without even passing the request to hydra
<gchristensen> anyone know how to do that?
<gchristensen> aanderse:
bennofs has quit [Remote host closed the connection]
bennofs has joined #nixos-dev
<supersandro2000> RewriteEngine On
<supersandro2000> RewriteRule ^/search?query=pkgs.chromium$ - [L,R=301,NC]
<supersandro2000> *RewriteRule ^/search?query=pkgs.chromium$ - [L,R=429,NC]
<supersandro2000> didn't test. exact match, maybe only begins with? flags in [] mean: last, redirect 429, nocase.
AlwaysLivid has joined #nixos-dev
<aanderse> yeah that looks reasonable 👍
LnL has joined #nixos-dev
kalbasit__ has joined #nixos-dev
WilliButz has joined #nixos-dev
mkaito has joined #nixos-dev
mkaito has joined #nixos-dev
kalbasit__ has quit [Ping timeout: 240 seconds]
<andi-> Did I miss that someone is constantly scraping for chromium?
<gchristensen> yeah, someone is
WilliButz has quit [Remote host closed the connection]
<gchristensen> supersandro2000: so RewriteRule ^/search?query=pkgs.chromium$ - [L,R=429,NC] made the "h" optional, so I escaped it as RewriteRule ^/search\?query=pkgs.chromium$ - [L,R=429,NC] and that doesn't match either. what do you suggest?
<symphorien[m]> `^/search.query=pkgs.chromium$`
WilliButz has joined #nixos-dev
<jonge[m]> hi there, i am interested in getting static c++ apps with clang. while playing with `makeStaticBinaries`, i found the following behavior: if i add `-static -lstdc++ -lfoo -lstdc++ -lbar -lstdc++` to `NIX_CFLAGS_LINK`, and run with `NIX_DEBUG=1`, i see that it translates those `-lstdc++` args to `-lc++`, as in `-static -lc++ -lfoo -lc++ -lbar -lc++`. i don't understand *where* in the wrappers this happens, can anybody
<jonge[m]> point me to the right position?
<supersandro2000> oh yeah maybe queries are special. As I said didn't test
<supersandro2000> also can we invalidate all ofborg results? I merged stdenv.lib got in master and it is the third PR today I did to fix such issue
WilliButz has quit [Quit: bye]
WilliButz has joined #nixos-dev
bennofs has quit [Quit: https://quassel-irc.org - Chat comfortably. Anywhere.]
<jonge[m]> infinisil: thank you. i just discovered that when clang++ is invoked with both `-stdlib=libc++` and `-lstdc++`, then it will translate `-lstdc++` to `-lc++`, but not because of anything in nixpkgs.
eyJhb has quit [Quit: Clever message]
eyJhb has joined #nixos-dev
eyJhb has joined #nixos-dev
<samueldr> I'll open an issue too, but I wrote a small reproduction for what I suppose is a bug with kernels and overrideAttrs
<samueldr> related to this earlier comment of mine I suppose: [02:47:57] <samueldr> my head hurts https://github.com/NixOS/nixpkgs/blob/21a2aebf314c385ac44c09492c76c31f846ae3e3/pkgs/os-specific/linux/kernel/generic.nix#L179-L187
<samueldr> I guess my first question is: should it just work? I suppose so, since for other derivations than the kernel, it just works
bennofs has joined #nixos-dev
<samueldr> updated gist with the workaround
<gchristensen> symphorien[m]: clever :P
AlwaysLivid has quit [Ping timeout: 268 seconds]
<{^_^}> resolved: RootPartitionLowDiskSpace: https://monitoring.nixos.org/prometheus/alerts
stolyaroleh_ has quit [Ping timeout: 265 seconds]
tilpner has quit [Remote host closed the connection]
<samueldr> #111504 for anyone curious about this weird issue with overrideAttrs and kernel derivations
<{^_^}> https://github.com/NixOS/nixpkgs/issues/111504 (by samueldr, 12 minutes ago, open): overrideAttrs on kernel derivation loses passthru values
tilpner has joined #nixos-dev
orivej has quit [Ping timeout: 264 seconds]
<infinisil> samueldr: (commented)
<samueldr> infinisil: Ah, I was confused
<samueldr> I thought you meant in _my_ code using `old` instead of expanding in a set-pattern function
<samueldr> and I wondered how the heck that could fix it
<infinisil> Ahh xD
<samueldr> yeah, sounds about right from my initial gut feeling
<samueldr> I wonder what that could break though
<samueldr> I mean, I definitely don't grok those bits, even though I was fully expecting they were at fault
<samueldr> infinisil: I assume the workaround still works with that fix applied, right?
* samueldr will test soon
<infinisil> Yeah
<samueldr> good
<samueldr> I'll be using the workaround for the time being where it matters
<samueldr> I might have rethought it otherwise
<{^_^}> firing: RootPartitionLowDiskSpace: https://monitoring.nixos.org/prometheus/alerts
cole-h has joined #nixos-dev
tokudan has joined #nixos-dev
<samueldr> infinisil++ that succint explanation of the issue
<{^_^}> infinisil's karma got increased to 409
<infinisil> :)
<samueldr> also nice to see it's pretty much what I thought was happening, but since I didn't know the details I couldn't really prove it