<simpson>
Okay, it looks like I can build "igraph", with C linkage, in its own Nix expression and with standard autotools, and then I should be able to pass that into the Python package. Great.
Sonarpulse has quit [Ping timeout: 245 seconds]
phreedom has quit [Remote host closed the connection]
phreedom has joined #nixos-dev
<thoughtpolice>
simpson: Eh, I'm guessing they probably package all the C++ code for Z3 with the python package, from PyPi, and if you install it that way it will just build everything for you. Looking at the setup.py in the PyPi package, this seems to be the case
<thoughtpolice>
simpson: What we should probably do instead, is have the existing Z3 package create a compatible Python output, one that you can use directly. Then we can just set `pythonPackages.z3-solver = z3.python;`, and we'll only ever need one Z3 derivation.
<simpson>
thoughtpolice: That appears to be the case. There's at least two problems with your approach; the first is that I happen to need `python3Packages.z3-solver` so maybe we need to build Z3 multiple times with multiple Pythons?
<simpson>
...Actually that's both problems and I'm bad at words.
<thoughtpolice>
simpson: I think we might need to compile the *bindings* twice, but not Z3 necessarily. With some finesse I believe it can be setup to do this cleanly. To be fair -- I am the Z3 maintainer, but I haven't used its Python bindings (most things just use smtlib2 I/O to communicate, IME...)
<simpson>
thoughtpolice: Yeah, and I'd be happy to not have to involve the existing Z3 setup, but any advice is welcome since it appears that the PyPI version doesn't Just Work.
<thoughtpolice>
It's possible things might not work out this cleanly; but IIRC, Python's Z3 bindings use a separate libz3_python.so that's linked to libz3.so, and the python FFI code basically just talks to the public API. I don't think it digs into anything very deep. So I think it should be possible to just build the bindings, once against python3Packages and once against python2
<thoughtpolice>
But yes, I'd have to look at it a bit...
<thoughtpolice>
simpson: There may already be a bug about python Z3 bindings, but -- if you could file a ticket and assign/CC me on it, that would be helpful, so I can keep up with it.
<simpson>
Well, no rush. I'm currently kicking igraph.
<gchristensen>
chaker: I think it probably shouldn't try an infinite number of times?
<chaker>
Yeah gchristensen, make sense. Well I think we can add a column in the database to check the number of times each notification is retried and limit by that. I will check if I can do something like that without changing the schema of the database.
<niksnut>
I don't think it needs to be stored in the database, in memory should be enough
<chaker>
Yep, I will do that. Thanks guys!
<grantwu>
peti: That's unfortunate. Doesn't that mean that locale support is broken for all Nix users not on NixOS?
pie_ has quit [Read error: Connection reset by peer]
pie_ has joined #nixos-dev
<peti>
grantwu: Not for all of them. Just for those who mix different versions of glibc on the same system.
<grantwu>
peti: How do I tell if I'm doing that?
<grantwu>
Hrm, I believe I am
<grantwu>
Er... so doesn't that mean "For everyone using Nix not on NixOS, except for those who happen to be on a distro that has the same glibc as Nix"?
goibhniu has quit [Ping timeout: 268 seconds]
<peti>
grantwu: There are two dimensions: what glibc version does your host system have and what glibc version does the Nixpkgs version have that you use.
goibhniu has joined #nixos-dev
<grantwu>
So if they match, no problem. If they don't match, problems. Right?
goibhniu has quit [Ping timeout: 264 seconds]
<grantwu>
Isn't nixpkgs-unstable the default for Nix on non-NixOS?
goibhniu has joined #nixos-dev
<peti>
grantwu: What do you mean by "default"? There's a number of different branches and you can choose whichever you want.
pie_ has quit [Ping timeout: 255 seconds]
<grantwu>
The fact that you can select channels does not imply that there is no default.
<grantwu>
nixpkgs-unstable is the default one selected. I just confirmed in #nixos.
obadz- has joined #nixos-dev
obadz has quit [Ping timeout: 260 seconds]
obadz- is now known as obadz
__Sander__ has quit [Ping timeout: 245 seconds]
__Sander__ has joined #nixos-dev
pie_ has joined #nixos-dev
orivej has joined #nixos-dev
pie_ has quit [Ping timeout: 240 seconds]
obadz- has joined #nixos-dev
obadz has quit [Ping timeout: 256 seconds]
obadz- is now known as obadz
chaker` has joined #nixos-dev
chaker has quit [Ping timeout: 265 seconds]
pie_ has joined #nixos-dev
matthewbauer has joined #nixos-dev
orivej has quit [Ping timeout: 268 seconds]
Sonarpulse has joined #nixos-dev
<peti>
grantwu: Ah, right. The Nix install script registers that channel by default. I forgot.
<dtz>
shlevy: (or anyone else): are we still using stock glibc 2.27? How were those patch gzip files created for 2.26? Lots of little fixes --possibly some security issues as well, dunno impact but changelog (glibc-2.27..origin/release/2.27/master) mentions a few CVE's.
__Sander__ has quit [Quit: Konversation terminated!]
matthewbauer has quit [Read error: Connection reset by peer]
matthewbauer has joined #nixos-dev
pie_ has joined #nixos-dev
<shlevy>
dtz: I don't know, I haven't applied any security patches
goibhniu has quit [Ping timeout: 256 seconds]
<dtz>
haha okay, I'll track down who that was sorry :)
<dtz>
vcunat: looks like you're the glibc-security-patches guy? :D :)
orivej has joined #nixos-dev
<andi->
dtz: iirc those were generate just by doing a git diff from in between the revisions we need.. E.g. release + 10 commits until release+N commits
<andi->
I did those a year ago.. Wasnt pleasant. I think in the git log the commands are mentioned once
<dtz>
haha okay, ty.
Cale has quit [Ping timeout: 255 seconds]
matthewbauer has quit [Ping timeout: 264 seconds]
Cale has joined #nixos-dev
matthewbauer has joined #nixos-dev
chaker` has quit [Ping timeout: 256 seconds]
phreedom has quit [Remote host closed the connection]
phreedom has joined #nixos-dev
matthewbauer has quit [Read error: Connection reset by peer]
matthewbauer has joined #nixos-dev
matthewbauer has quit [Read error: Connection reset by peer]
matthewbauer has joined #nixos-dev
matthewbauer has quit [Read error: Connection reset by peer]
matthewbauer has joined #nixos-dev
<aminechikhaoui>
hm question, what's the reasonable behavior of an expression like http://termbin.com/t0ry ?
<aminechikhaoui>
I guess nix-build hanging with that is sort of a bug