sage: 8.0 -> 8.1, fix sandbox build #34291
Conversation
The sandboxed build was failing, because it relied heavily on /usr/bin/env. This is fixed with a lot of shebang patching (both to system packages and to sage-internal packages).
|
@GrahamcOfBorg build sage |
|
After looking at what is needed for a rebuild-everything-locally sandboxed build, I have gone into negative-amount-of-surprise about my failure to build it while reusing maxima-ecl etc. built externally… |
GrahamcOfBorg
added
10.rebuild-darwin: 0
|
Yeah, sage is a real pain! Lets just hope they don't change their build too much in the future. |
There was a problem hiding this comment.
Failure for system: x86_64-darwin
Package ‘sage-8.1’ in /Users/graham/nix-borg/repo/38dca4e3aa6bca43ea96d2fcc04e8229/builder/grahamc-zoidberg/pkgs/applications/science/math/sage/default.nix:195 is not supported on ‘x86_64-darwin’, refusing to evaluate.
a) For `nixos-rebuild` you can set
{ nixpkgs.config.allowBroken = true; }
in configuration.nix to override this.
b) For `nix-env`, `nix-build`, `nix-shell` or any other Nix command you can add
{ allowBroken = true; }
to ~/.config/nixpkgs/config.nix.
|
An optimist would hope that they would actually change their build in the future and make it easier to package with system versions of software! Oh well… On the other hand, Debian does seem to be forcing Sage to reuse at least some of the dependencies from the outer system. |
There was a problem hiding this comment.
Failure for system: x86_64-linux
[openblas-0.2.20.p1] Warning: 'scale2' may be used uninitialized in this function [-Wmaybe-uninitialized]
[openblas-0.2.20.p1] gfortran -O2 -Wall -m64 -fPIC -c -o stpcon.o stpcon.f
[openblas-0.2.20.p1] gfortran -O2 -Wall -m64 -fPIC -c -o stprfs.o stprfs.f
[openblas-0.2.20.p1] gfortran -O2 -Wall -m64 -fPIC -c -o stptri.o stptri.f
[openblas-0.2.20.p1] stptri.f:229:0:
[openblas-0.2.20.p1]
[openblas-0.2.20.p1] $ AP( JCLAST ), AP( JC+1 ), 1 )
[openblas-0.2.20.p1]
building of ‘/nix/store/my9pnv0yd9wdblk0lzjk0w49rik53jax-sage-8.1.drv’ timed out after 1800 seconds
error: build of ‘/nix/store/my9pnv0yd9wdblk0lzjk0w49rik53jax-sage-8.1.drv’ failed
|
@grahamc timeout means no output for half an hour? Any way to see if there was a sudden load on the builder machine? |
|
The timeout is for overall build time. |
|
There goes my plan to improve builder utilisation… |
|
(actually tested on my sandbox-enabled build brix) Thanks! |
|
Great, thanks! |
|
This commit is included in |
|
I… don't understand what has happenned. https://hydra.nixos.org/job/nixpkgs/trunk/sage.x86_64-linux |
|
Well thats an easy fix, but how did that happen? Can it be that |
|
I am not really sure. |
|
I opened an issue (#34387) about it, hopefully somebody more knowledgable about the sandbox and texlive-combined can explain and maybe fix the issue. Its probably better to wait for that before I make a PR for a fix, since there may be more latex packages missing and I don't want to abuse hydra as a test machine. |
This updates sage to 8.1 and, more importantly, makes the build work sandboxed.
Motivation for this change
The sandboxed build was failing, because it relied heavily on /usr/bin/env. This is fixed with a lot of shebang patching (both to system packages and to sage-internal packages).
Things done
build-use-sandboxinnix.confon non-NixOS)nix-shell -p nox --run "nox-review wip"./result/bin/) (Well, there are too many to test all. But at least ./result/bin/sage)