<lukegb>
Am I missing something blindingly obvious? python3.pkgs.pyqt5 has things in its propagatedBuildInputs, but no /nix-support/propagated-build-inputs in its main o... oh, it's in the dev one
<samueldr>
"yes" I assume is the answer
<samueldr>
(but really maybe it's not?)
<lukegb>
Interestingly that means this software doesn't work if you put "pyqt5" in its pythonPath
<lukegb>
but _does_ work if you put "pyqt5.dev" in its pythonPath
<sterni>
depends on put it in its python path how probably
<rnhmjoj>
anyone familiar with cmake willing to help me? i'm stuck trying to replace a bundled library with the one from nixpkgs in a package
<rnhmjoj>
this is what i'm doing: https://maxwell.ydns.eu/~rnhmjoj/monero.patch the library is found by cmake and it builds, but while linking the linker complains about missing symbols from the library
<siraben>
cmake is scary, I cry every time something cursed happens while packaging a cmake-based package
<rnhmjoj>
yeah, on every minor or bugfix update this thing manages to break the build
<jtojnar>
rnhmjoj: maybe move it from private to public?
<jtojnar>
that way it should get inherited from qrdecoder by whatever links against qrdecoder
<jtojnar>
zhaofeng: I would expect those to be much more upstreamable so it would probably be fine to include them in the regular packages
<jtojnar>
would need to evaluate the individual patches to be sure
<rnhmjoj>
Jan Tojnar: the problem seems to be the library is not even being linked to qrcodereader (if i make it SHARED and run ldd on it, it's not there)
<jtojnar>
rnhmjoj: it will probably depend on how the library is being linked to qrcodereader
<jtojnar>
unfortunately, with this ancient-style cmake code you need to call multiple different functions for it to work properly
<rnhmjoj>
it's ancient? it's funny because they just moved from a qmake-based build system and this is supposed to be "modern"
<jtojnar>
rnhmjoj: they might actually be using modern CMake (they do it for Qt) but I have not seen the rest of the code
<jtojnar>
most projects unfortunatelly use CMake 2 idioms (sometimes with few instances of modern CMake to rub the salt into the wound)
<rnhmjoj>
tried all combinations of STATIC/SHARED PRIVATE/PUBLIC but still not being linked
<siraben>
I just revisited it after 2 months and still doesn't work :(
contrapumpkin has joined #nixos-dev
copumpkin has quit [Ping timeout: 252 seconds]
copumpkin has joined #nixos-dev
contrapumpkin has quit [Ping timeout: 268 seconds]
<qyliss>
zhaofeng: maybe you could have a "purism" attribute, and then group whatever is needed under that? e.g. purism.gtk3. Would make them easy to all find, and keep them all in one place.
__monty__ has joined #nixos-dev
jonringer has quit [Ping timeout: 276 seconds]
<jtojnar>
siraben: instead of autoreconfhook run ./bootstap manually in preConfigure
<rnhmjoj>
Jan Tojnar: turns out the version from nixpkgs wasn't ABI compatible and needed a bump. anyway, thank you for the sanity check
<qyliss>
I use a VPS that's run on Xen, so I know about it from the guest's side, but that sort of thing is taken care of my the VPS host
<samueldr>
qyliss: thanks
jonringer has joined #nixos-dev
mkaito has joined #nixos-dev
mkaito has quit [Client Quit]
mkaito has joined #nixos-dev
<lukegb>
fyi: if you're relying on the .ls files in the nixos cache, ones uploaded in the last month or so may be invalid JSON (and may be orders of magnitude larger than expected, in some cases - ~40MB -> ~4GB)
<lukegb>
it'll get "fixed" whenever the next world rebuild happens
mkaito has quit [Quit: WeeChat 3.1]
orivej has quit [Ping timeout: 240 seconds]
pingiun has joined #nixos-dev
dongcarl has joined #nixos-dev
<samueldr>
oh no :( make-disk-image shouldn't be expanded upon anymore :(
<samueldr>
it's all the wrong abstractions you could make :/
<gchristensen>
yeah, just got email about thath :)
<samueldr>
we need to declare filesystems builders / disk image builders bankruptcy!
<gchristensen>
+10
<gchristensen>
let's do it, once and for all, and tear down these old todo's from copumpkin from a decade ago
<samueldr>
my time's a bit lacking for that :/
<samueldr>
and, I say that, but I don't mean to imply my proposal is the right way to go
<samueldr>
(it is, though, I think)
<samueldr>
maybe it warrants something like a CFP? call for proposals, from any interested parties? kind of flipping the RFC process on its head, where we have identified an issue, but want opinions about how to go forward?
<samueldr>
since, in **all** cases here, it's a big chunk of changes
<gchristensen>
I like that idea
<gchristensen>
but I also think putting yours forward initially is useful
<gchristensen>
might inspire alterantives :)
<qyliss>
CFPs sound like a good idea in general
<MichaelRaskin>
CFP should be an RFC with requirements establishing a commitment to accept any design satisfying the requirements
__monty__ has quit [Quit: leaving]
copumpkin has joined #nixos-dev
<lukegb>
I wondered why the NixOS tests were running _super slowly_. Turns out if you, err, try to run them in a VM without nested virt then that's kinda what's gonna happen :P
rj has quit [Ping timeout: 240 seconds]
contrapumpkin has joined #nixos-dev
copumpkin has quit [Ping timeout: 252 seconds]
contrapumpkin is now known as copumpkin
supersandro2000 has quit [Killed (verne.freenode.net (Nickname regained by services))]