openblas: 0.3.4 -> 0.3.5, rework a bit #53972
Conversation
|
This needs a rebase after my CC change. |
* doCheck should be 'true' in (at least) the non-cross case, this looks like an inverted check that's largely benign * doCheck will be set to 'false' in the cross case anyway, makeDerivation does this IIRC * targetPrefix can be used without checking, probably by design Derivation hash does change but no "real" functionality change intended.
(but set appropriately for cross and non-cross cases both) * I'm not sure what NO_BINARY_MODE does, this change now sets explicitly false in the non-cross scenario (previously unset unless cross). * Drop musl NO_AFFINITY case, will be removed in upgrade shortly
dtzWill
force-pushed
the
update/openblas-0.3.5
branch
from
f40d37f
to
4738517
Compare
|
rebased! Basically same changes but across a few commits to make easier for review, squash as preferred (or LMK and i'll do so). |
|
Thanks for the improvements and the |
|
FWIW #54886 conflicts with this, I resolved it in my tree with this result: https://github.com/dtzWill/nixpkgs/blob/1c5bfc30a7fcf22138adb8f8441d2c40dda201e1/pkgs/development/libraries/science/math/openblas/default.nix Quick notes about what was done and why: dtzWill@1c5bfc3 Thought I'd share for whoever/whenever master and staging get merged. |
|
This commit (1f046700) broke openblas build on aarch64-linux. (I re-tried the commit itself and its parent.) Error logs e.g. on Hydra: https://hydra.nixos.org/build/87629037 |
|
Is it related to the |
|
Hmm, two things I noticed:
Very sorry re:1, although it's a shame because I like this much better. But broken is worse than ugly! :P |
|
Note: I just merged this into master via c40f211. I didn't feel like attempting to solve this as well, and I guess blas on aarch64 isn't such a big deal, so we might afford to break it temporarily (I hope). |
|
BTW, this darwin error also seems new: https://nix-cache.s3.amazonaws.com/log/agn0dzc87gb4mnrf3kyswr30zbr92kmm-openblas-0.3.5.drv |
My earlier change mistakenly expected `toString false` to produce '0' instead of the empty string, leading to unexpected config changes. Intended to address issue mentioned here and in following discussion: NixOS#53972 (comment) Sorry, folks! (special-case handling of bools here makes this "cleanup" a bit less of an obvious win but hopefully still preferable overall :)) ----------- makeFlags in resulting derivation, according to this one-liner: $ nix show-derivation -f . openblas|jq ".[].env.makeFlags" before: "BINARY=64 CC=cc CROSS= DYNAMIC_ARCH=1 FC=gfortran HOSTCC=cc INTERFACE64=1 NO_BINARY_MODE= NO_STATIC=1 NUM_THREADS=64 PREFIX=/1rz4g4znpzjwh1xymhjpm42vipw92pr73vdgl6xs1hycac8kf2n9 TARGET=ATHLON USE_OPENMP=1" after: "BINARY=64 CC=cc CROSS=0 DYNAMIC_ARCH=1 FC=gfortran HOSTCC=cc INTERFACE64=1 NO_BINARY_MODE=0 NO_STATIC=1 NUM_THREADS=64 PREFIX=/1rz4g4znpzjwh1xymhjpm42vipw92pr73vdgl6xs1hycac8kf2n9 TARGET=ATHLON USE_OPENMP=1" Without knowing how `placeholder` works, it seems interesting if entirely unrelated that the `PREFIX` is same for both! :). TIL.
My earlier change mistakenly expected `toString false` to produce '0' instead of the empty string, leading to unexpected config changes. Intended to address issue mentioned here and in following discussion: NixOS#53972 (comment) Sorry, folks! (special-case handling of bools here makes this "cleanup" a bit less of an obvious win but hopefully still preferable overall :)) ----------- makeFlags in resulting derivation, according to this one-liner: $ nix show-derivation -f . openblas|jq ".[].env.makeFlags" before: "BINARY=64 CC=cc CROSS= DYNAMIC_ARCH=1 FC=gfortran HOSTCC=cc INTERFACE64=1 NO_BINARY_MODE= NO_STATIC=1 NUM_THREADS=64 PREFIX=/1rz4g4znpzjwh1xymhjpm42vipw92pr73vdgl6xs1hycac8kf2n9 TARGET=ATHLON USE_OPENMP=1" after: "BINARY=64 CC=cc CROSS=0 DYNAMIC_ARCH=1 FC=gfortran HOSTCC=cc INTERFACE64=1 NO_BINARY_MODE=0 NO_STATIC=1 NUM_THREADS=64 PREFIX=/1rz4g4znpzjwh1xymhjpm42vipw92pr73vdgl6xs1hycac8kf2n9 TARGET=ATHLON USE_OPENMP=1" Without knowing how `placeholder` works, it seems interesting if entirely unrelated that the `PREFIX` is same for both! :). TIL.
My earlier change mistakenly expected `toString false` to produce '0' instead of the empty string, leading to unexpected config changes. Intended to address issue mentioned here and in following discussion: #53972 (comment) Sorry, folks! (special-case handling of bools here makes this "cleanup" a bit less of an obvious win but hopefully still preferable overall :)) ----------- makeFlags in resulting derivation, according to this one-liner: $ nix show-derivation -f . openblas|jq ".[].env.makeFlags" before: "BINARY=64 CC=cc CROSS= DYNAMIC_ARCH=1 FC=gfortran HOSTCC=cc INTERFACE64=1 NO_BINARY_MODE= NO_STATIC=1 NUM_THREADS=64 PREFIX=/1rz4g4znpzjwh1xymhjpm42vipw92pr73vdgl6xs1hycac8kf2n9 TARGET=ATHLON USE_OPENMP=1" after: "BINARY=64 CC=cc CROSS=0 DYNAMIC_ARCH=1 FC=gfortran HOSTCC=cc INTERFACE64=1 NO_BINARY_MODE=0 NO_STATIC=1 NUM_THREADS=64 PREFIX=/1rz4g4znpzjwh1xymhjpm42vipw92pr73vdgl6xs1hycac8kf2n9 TARGET=ATHLON USE_OPENMP=1" Without knowing how `placeholder` works, it seems interesting if entirely unrelated that the `PREFIX` is same for both! :). TIL.
Motivation for this change
https://github.com/xianyi/OpenBLAS/releases/tag/v0.3.5
Things done
sandboxinnix.confon non-NixOS)nix-shell -p nox --run "nox-review wip"./result/bin/)nix path-info -Sbefore and after)