glpk: adopt error recovery patch (#44201)

This makes it possible to use "vanilla" glpk for sage and shouldn't
affect anything else.
This commit is contained in:
Timo Kaufmann
2018-07-29 20:53:56 +02:00
committed by xeji
parent 2791900608
commit ae821e09e7
2 changed files with 22 additions and 36 deletions
@@ -12,8 +12,6 @@ let
cysignals = super.cysignals.override { inherit pari; };
cvxopt = super.cvxopt.override { inherit glpk; };
# python packages that appear unmaintained and were not accepted into the nixpkgs
# tree because of that. These packages are only dependencies of the more-or-less
# deprecated sagenb. However sagenb is still a default dependency and the doctests
@@ -26,7 +24,7 @@ let
pybrial = self.callPackage ./pybrial.nix {};
sagelib = self.callPackage ./sagelib.nix {
inherit flint ecl pari glpk eclib ntl arb;
inherit flint ecl pari eclib ntl arb;
inherit sage-src openblas-blas-pc openblas-cblas-pc openblas-lapack-pc pynac singular;
linbox = nixpkgs.linbox.override { withSage = true; };
};
@@ -47,7 +45,7 @@ let
};
sage-env = self.callPackage ./sage-env.nix {
inherit sage-src python rWrapper openblas-cblas-pc glpk ecl singular eclib pari palp flint pynac pythonEnv giac ntl;
inherit sage-src python rWrapper openblas-cblas-pc ecl singular eclib pari palp flint pynac pythonEnv giac ntl;
pkg-config = nixpkgs.pkgconfig; # not to confuse with pythonPackages.pkgconfig
};
@@ -176,37 +174,6 @@ let
sha256 = "19gbsm8jqq3hraanbmsvzkbh88iwlqbckzbnga3y76r7k42akn7m";
};
});
# https://trac.sagemath.org/ticket/24824
glpk = nixpkgs.glpk.overrideAttrs (attrs: rec {
version = "4.63";
name = "glpk-${version}";
src = fetchurl {
url = "mirror://gnu/glpk/${name}.tar.gz";
sha256 = "1xp7nclmp8inp20968bvvfcwmz3mz03sbm0v3yjz8aqwlpqjfkci";
};
patches = (attrs.patches or []) ++ [
# Alternatively patch sage with debians
# https://sources.debian.org/data/main/s/sagemath/8.1-7/debian/patches/t-version-glpk-4.60-extra-hack-fixes.patch
# The header of that debian patch contains a good description of the issue. The gist of it:
# > If GLPK in Sage causes one error, and this is caught by Sage and recovered from, then
# > later (because upstream GLPK does not clear the "error" flag) Sage will append
# > all subsequent terminal output of GLPK into the error_message string but not
# > actually forward it to the user's terminal. This breaks some doctests.
(fetchpatch {
name = "error_recovery.patch";
url = "https://git.sagemath.org/sage.git/plain/build/pkgs/glpk/patches/error_recovery.patch?id=07d6c37d18811e2b377a9689790a7c5e24da16ba";
sha256 = "0z99z9gd31apb6x5n5n26411qzx0ma3s6dnznc4x61x86bhq31qf";
})
# Allow setting a exact verbosity level (OFF|ERR|ON|ALL|DBG)
(fetchpatch {
name = "exact_verbosity.patch";
url = "https://git.sagemath.org/sage.git/plain/build/pkgs/glpk/patches/glp_exact_verbosity.patch?id=07d6c37d18811e2b377a9689790a7c5e24da16ba";
sha256 = "15gm5i2alqla3m463i1qq6jx6c0ns6lip7njvbhp37pgxg4s9hx8";
})
];
});
in
python.pkgs.sage-wrapper // {
doc = python.pkgs.sagedoc;