<colemickens>
gchristensen: do you use non-1.0 scaling anywhere? And if so, have you found the correct incantation for Qt apps? It seems a recent Qt bump changed things and I'm having a heck of a time getting Qt to look right in any scenario (laptop, mixed laptop/external, etc)
<colemickens>
("recent" being some months ago in nixpkgs but a recent Qt release anyway)
<ivan>
I have just a dpi configured and it seems to work fine
<ivan>
do you have extra scaling factors somewhere?
emily has quit [Ping timeout: 250 seconds]
<colemickens>
sorry, I'm specifically using Sway and DPI in Wayland is different. not sure if you are as well?
<colemickens>
I don't really have anything set anywhere, other than having configured the scaling factors for the monitors in sway's config.
<colemickens>
But I did try every combination of envvars supposedly respected by this version of QT that is meant to tweak its scaling behavior and never really did notice a positive improvement
emily has joined #nixos-chat
lopsided98 has quit [Quit: Disconnected]
lopsided98 has joined #nixos-chat
iqubic has joined #nixos-chat
emily has quit [Quit: Updating details, brb]
emily has joined #nixos-chat
drakonis has quit [Quit: WeeChat 2.4]
lassulus has quit [Ping timeout: 255 seconds]
lassulus has joined #nixos-chat
endformationage has quit [Ping timeout: 246 seconds]
Myhlamaeus has quit [Ping timeout: 252 seconds]
__monty__ has joined #nixos-chat
<eyJhb>
Need a Mac OS X expert. Can Mac boot a normalt Ubuntu Live USB, even if not formatted with their Journal stuff?
<__monty__>
Booting is pre-OS so mac OS has nothing to do with it.
<__monty__>
The problem's probably their uefi setup.
<__monty__>
Macs can definitely read UDF though, which is the most common (only?) filesystem used on CDs.
<__monty__>
So maybe the live images are UDF formatted?
<__monty__>
(Side note: if you're ever considering formatting something FAT to use it cross platform, consider UDF instead works like a charm and doesn't come with FAT's annoying limitations/incarnations.
<joepie91>
__monty__: CD-ROMs generally use ISO9660, DVDs either ISO9660 or UDF, I think Blu-Rays generally use UDF
<ar>
__monty__: iirc their efi doesn't support fat for ESP
<__monty__>
joepie91: I was gonna just say disks but didn't want the HDD/SSD confusion *and* I'm old which is why I thought of CD before DVD didn't really specifically mean CDs, thanks for the clarification though.
<__monty__>
ar: Good to know. Another reason to recommend against FAT.
<ar>
__monty__: the problem is, everything else standarized on using fat for ESP
<__monty__>
But FAT should die. Preferably in a fire.
<ar>
but it won't die, because everyone and their dogs uses it
<__monty__>
So you're saying the problem is dog people?
<ar>
… no
<__monty__>
Oh, come on, grumpy cat needs to be avenged! ; )
<MichaelRaskin>
can we please kill exFAT first?
<sphalerite>
eyJhb: depends on the hardware version, older intel macs should work fine
<ar>
MichaelRaskin: no, because that's what cameras use these days
<MichaelRaskin>
Which is exactly the problem, why even bother killing FAT if nothing uses it
<yorick>
but we need the FAT for the UEFI
<joepie91>
__monty__: btw, I was wrong, there *is* a communities feature in Matrix
<tilpner>
"We have previously attempted to implement Communities but we are dissatisfied with the user experience (we also have open questions on the API) and want to take another swing at it."
<tilpner>
Groups as rooms might make it easier to join groups without groups-specific client support
gchristensen has joined #nixos-chat
{^_^} has joined #nixos-chat
{^_^} has quit [Remote host closed the connection]
{^_^} has joined #nixos-chat
pie_ has joined #nixos-chat
ivan has quit [Quit: lp0 on fire]
<infinisil>
Ohh, Discourse has great vim-style keybinding integration :O
<infinisil>
Well, integration, it just supports vim-style commands
<infinisil>
Very nice with scrolling
ivan has joined #nixos-chat
__monty__ has quit [Quit: leaving]
pie__ has joined #nixos-chat
pie_ has quit [Ping timeout: 252 seconds]
<Shados>
yorick: There are UEFI implementations that support non-FAT ESPs. And, of course, if we all had coreboot+tianocore-compatible systems we could compile in EFI drivers for whatever filesystems we liked...
<ar>
Shados: any other than the one on macs that supports hfs?
<Shados>
ar: Not off the top of my head, although there's *something* niggling at me...
endformationage has joined #nixos-chat
drakonis has joined #nixos-chat
drakonis1 has joined #nixos-chat
drakonis_ has quit [Ping timeout: 244 seconds]
drakonis has quit [Ping timeout: 250 seconds]
avn_ has joined #nixos-chat
nckx has quit [Read error: Connection reset by peer]
nckx has joined #nixos-chat
pie__ has quit [Ping timeout: 258 seconds]
<samueldr>
it might be possible to simply add the drivers to an existing uefi bios, flash the new one, and cross fingers https://efi.akeo.ie/
<samueldr>
will an existing implementation pick an EF00 partition on another FS than FAT? no way to know without testing (or disassembling and readinc impl.) I guess
<MichaelRaskin>
Fortunately you can test on an external drive in most cases
<samueldr>
you mean having an EF00 on another drive, yeah most likely, but still need to flash the bios/firmware :)
<samueldr>
(well, unless you have a neat trick to share!)
<MichaelRaskin>
Depends — sometimes you can use Qemu
<samueldr>
hm, I would have assumed whatever bios shipped on my machine wouldn't like being on another different system, qemu included... but I guess qemu can be massaged into looking right enough
drakonis has joined #nixos-chat
drakonis1 has quit [Ping timeout: 250 seconds]
drakonis_ has joined #nixos-chat
drakonis has quit [Ping timeout: 252 seconds]
<nh2>
dtz: I'm having this trouble with libz. I'm trying to refactor static-haskell-nix to be made up of overlays on top of `pkgsMusl`. For any package that's not zlib, if I override it, that works. But for zlib, it completely skips what I'm declaring. Even if I do `zlib = 1`, which should fail, I can still ask for zlib and it returns the output.
<nh2>
As in `(pkgs.extend (final: previous: { zlib = 1; })).zlib` just returns `pkgs.zlib`, instead of my override.
<nh2>
I don't get what's going on there, and why it behaves that way only for zlib.
<Ralith>
where is pkgsMusl found?
<Ralith>
I've been wanting to use nix to build statically linked binaries for ages but nobody ever knows anything about it :/
<tilpner>
Ralith: Found, as in defined? stage.nix
<Ralith>
tilpner: as in referenced
<Ralith>
via
<Ralith>
how do you actually use it
<tilpner>
It's an alternative nixpkgs, like pkgsi686Linux
<tilpner>
You just use the package pkgsMusl.hello instead of hello
<tilpner>
And that may or may not build
<Ralith>
awesome
<tilpner>
Note that there's also pkgsStatic
<tilpner>
It's almost like pkgsMusl, but has an extra overlay
<tilpner>
It seems to enableStatic a few packages and use makeStaticBinaries
drakonis has joined #nixos-chat
<tilpner>
(See pkgs/top-level/static.nix)
<nh2>
Ralith: right now in nixpkgs master, and 19.03, you can do `nix-build -A pkgsStatic.hello` and you will obtain a statically linked GNU hello executable
<Ralith>
nh2: re: 'work on any linux distribution' -- doesn't the interpreter still need to be patched?
<Ralith>
oh, I guess static executables don't need an interpreter
<nh2>
Ralith: interpreter?
<nh2>
tilpner: do you understand the definition of `pkgsMusl` in `stage.nix`? I can understand `pkgsStatic`, that does `crossOverlays = [ (import ./static.nix) ];` and that's how its overrides come in. But `pkgsMusl`, what does it actually do? Where does it say that musl should be used?
<Ralith>
dynamic loader
<Ralith>
as in `patchelf --set-interpreter`
<nh2>
Ralith: yes, static exes don't have that
<Ralith>
awesome
<tilpner>
nh2: It sets a few ABIs, e.g. gnu = abis.musl
<nh2>
tilpner: but how does setting those ABIs resolve in eventually musl code being used?
<tilpner>
I'm not familiar with the intricacies of stdenv, but it must somehow use musl after being told to use the musl abi
<nh2>
hmm
<nh2>
I was hoping that learning that will eventually tell me why the zlib override doesn't work
* Ralith
tries rebuilding one of his projects with pkgsStatic, immediately collides headfirst with `anonymous function at /nix/store/.../libjpeg-turbo/default.nix:1:1 called with unexpected argument 'static'` :(
pie_ has joined #nixos-chat
<nh2>
Ralith: on 19.03 or master?
<Ralith>
19.03
<nh2>
In general master is better, as pkgsStatic is pretty new and I also found a couple things to fix in the last week
<Ralith>
good to know
<nh2>
libjpeg-turbo was one thing I also had problems with
<nh2>
though it was a linker error, not an evaluation error
<nh2>
ah, looks like the fact that I can't override zlib is not musl specific, also doesn't work for normal nixpkgs!
<tilpner>
nh2: FYI, overlays usually use self: super:, which appears to be the opposite of your naming
<tilpner>
nh2: Irrelevant here, but might help with some later problem
pie_ has quit [Remote host closed the connection]
pie_ has joined #nixos-chat
<nh2>
tilpner: oh, you are right, that is just a typo I made here. In my file I have `final: previous:`
<nh2>
thanks for noticing
<nh2>
I've fixed the github issue to avoid confusion
<tilpner>
nh2: Found the issue
<tilpner>
See bootstrap-stage4 in pkgs/stdenv/linux/default.nix
<tilpner>
Then compare with toFix from stage.nix
pie_ has quit [Ping timeout: 258 seconds]
<Ralith>
tilpner: wait, why does pkgsStatic.stdenv not have cc :|
<tilpner>
Huh
<Ralith>
to be specific: applying nix-shell to pkgsStatic.stdenv.mkDerivation gets me no cc
<Ralith>
or gcc, or clang, or whatever
<Ralith>
looks like it's x86_64-unknown-linux-musl-gcc
<Ralith>
the rust ecosystem does not expect `cc` to not exist
<tilpner>
nh2: Does that make sense to you?
<nh2>
But I'm not sure where to look at in "compare with toFix from stage.nix"
<nh2>
tilpner: the first yes, looks like the culprit is `inherit (prevStage) gettext gnum4 bison gmp perl texinfo zlib linuxHeaders;`, and none of those packages can be overridden?
<tilpner>
nh2: The toFix function has stdenvOverrides after overlays
<tilpner>
And uses that last while folding with extend
<tilpner>
# The complete chain of package set builders, applied from top to bottom.
<tilpner>
# stdenvOverlays must be last as it brings package forward from the
<tilpner>
# previous bootstrapping phases which have already been overlayed.
<tilpner>
It even says that this working-as-intended™
<nh2>
tilpner: ah, I see. So it looks like you found the reason, but we don't have a solution yet as overriding zlib would be very useful. Do you want to comment on the issue with your findings, or shall I do it?
<tilpner>
Go ahead
<Ralith>
got my project to build and it's still ending up with dynamically linked libc, libstdc++, and libgcc_s
<Ralith>
looks like more rust issues
<lucus16>
How are pull requests that need to rebuild the world handled?
<infinisil>
lucus16: Those should go to staging
<lucus16>
As in they should use base branch staging?
<infinisil>
Yup
pie_ has joined #nixos-chat
_e has quit [Quit: WeeChat 2.4]
_e has joined #nixos-chat
_e has quit [Client Quit]
_e has joined #nixos-chat
<samueldr>
gimp has some flair in its standard output "EEEEeEeek! 1 GeglBuffers leaked"
gchristensen has quit [Quit: Connection closed for inactivity]
Myhlamaeus has joined #nixos-chat
<drakonis>
so, why is nix so high up in the popular github repos list?
<drakonis>
nixpkgs that is
<Ralith>
shouldn't it be?
<samueldr>
drakonis: maybe (1) elaborate on why you think it should or shouldn't be and (2) link to where you see that, cannot seem to see that in trending or explore
<samueldr>
you seem surprised that it is though
<drakonis>
i'm surprised that it is so up high among other much bigger projects in terms of corporate investment
<drakonis>
this looks like it was done by a third party
<drakonis>
they link to a blockchain paper (?????)
<drakonis>
for the reputation algorithm
<samueldr>
you can order a list any way you want :) a list of github projects with a Levenshtein distance from nixpkgs would rank nixpkgs quite high!
<samueldr>
but yeah, not sure what reputation here means :/
<samueldr>
the linked yellow paper is a bit thick for me (or am I thick?)
<samueldr>
bleh, I can't find whether I'm misrembering a bit or not about GPT and shenanigans
<samueldr>
I have been successful at leaving a gap at the beginning of the disk (to embed u-boot), but I thought I would have been able to put a partition between the header and the table
<samueldr>
but cgpt tells me that those sectors are outside of the valid region; entries defined at sectors from start of disk and not from start of table
<samueldr>
maybe it's an implementation detail and the table always refers to space after itself?
<samueldr>
hm what I see seems to say "First LBA", and an LBA I guess is always addressed as an absolute location from start of disk, so hmm