Steps to reproduce
When building a new package which depends on py-numba spack concretizes to use the oldest available versionpy-numba@0.35.0 instead the newest one py-numba@0.55.1 without any reason. Therefore also downgrading python to version 3.6.15
Reproduce:
- fresh spack clone
- no config (removed
~/.spack)
- create a new package with a
py-numba dependency
I used py-fracridge (the package I wanted to create when encountering this problem) with the package.py
from spack import *
class PyFracridge(PythonPackage):
"""Fractional Ridge Regression."""
homepage = "https://nrdg.github.io/fracridge"
pypi = "fracridge/fracridge-1.4.3.tar.gz"
version('1.4.3', sha256='0446d486f00fea02110567fd9df14b8b2a7b155dc72700af9201873ea11c27cc')
depends_on('py-numba', type=('build', 'run'))
ping @adamjstewart @alalazo
Error message
$ spack solve py-fracridge
==> Best of 6 considered solutions.
==> Optimization Criteria:
Priority Criterion Installed ToBuild
1 number of packages to build (vs. reuse) - 30
2 deprecated versions used 0 0
3 version weight 0 0
4 number of non-default variants (roots) 0 0
5 preferred providers for roots 0 0
6 default values of variants not being used (roots) 0 0
7 number of non-default variants (non-roots) 0 0
8 preferred providers (non-roots) 0 0
9 compiler mismatches 0 0
10 OS mismatches 0 0
11 non-preferred OS's 0 0
12 version badness 0 71
13 default values of variants not being used (non-roots) 0 1
14 non-preferred compilers 0 0
15 target mismatches 0 0
16 non-preferred targets 0 0
py-fracridge@1.4.3%gcc@11.3.1 arch=linux-fedora35-skylake
^py-numba@0.35.0%gcc@11.3.1 arch=linux-fedora35-skylake
^py-numpy@1.19.5%gcc@11.3.1+blas+lapack patches=8a9d5d1 arch=linux-fedora35-skylake
^openblas@0.3.20%gcc@11.3.1~bignuma~consistent_fpcsr~ilp64+locking+pic+shared symbol_suffix=none threads=none arch=linux-fedora35-skylake
^perl@5.34.1%gcc@11.3.1+cpanm+shared+threads arch=linux-fedora35-skylake
^berkeley-db@18.1.40%gcc@11.3.1+cxx~docs+stl patches=b231fcc arch=linux-fedora35-skylake
^bzip2@1.0.8%gcc@11.3.1~debug~pic+shared arch=linux-fedora35-skylake
^diffutils@3.8%gcc@11.3.1 arch=linux-fedora35-skylake
^libiconv@1.16%gcc@11.3.1 libs=shared,static arch=linux-fedora35-skylake
^gdbm@1.19%gcc@11.3.1 arch=linux-fedora35-skylake
^readline@8.1%gcc@11.3.1 arch=linux-fedora35-skylake
^ncurses@6.2%gcc@11.3.1~symlinks+termlib abi=none arch=linux-fedora35-skylake
^pkgconf@1.8.0%gcc@11.3.1 arch=linux-fedora35-skylake
^zlib@1.2.12%gcc@11.3.1+optimize+pic+shared patches=0d38234 arch=linux-fedora35-skylake
^py-cython@0.29.24%gcc@11.3.1 arch=linux-fedora35-skylake
^py-pip@21.3.1%gcc@11.3.1 arch=linux-fedora35-skylake
^python@3.6.15%gcc@11.3.1+bz2+ctypes+dbm~debug+ensurepip+libxml2+lzma~nis~optimizations+pic+pyexpat+pythoncmd+readline+shared+sqlite3+ssl~tix~tkinter~ucs4+uuid+zlib arch=linux-fedora35-skylake
^expat@2.4.8%gcc@11.3.1+libbsd arch=linux-fedora35-skylake
^libbsd@0.11.5%gcc@11.3.1 arch=linux-fedora35-skylake
^libmd@1.0.4%gcc@11.3.1 arch=linux-fedora35-skylake
^gettext@0.21%gcc@11.3.1+bzip2+curses+git~libunistring+libxml2+tar+xz arch=linux-fedora35-skylake
^libxml2@2.9.13%gcc@11.3.1~python arch=linux-fedora35-skylake
^xz@5.2.5%gcc@11.3.1~pic libs=shared,static arch=linux-fedora35-skylake
^tar@1.34%gcc@11.3.1 arch=linux-fedora35-skylake
^libffi@3.4.2%gcc@11.3.1 arch=linux-fedora35-skylake
^openssl@1.1.1o%gcc@11.3.1~docs~shared certs=system arch=linux-fedora35-skylake
^sqlite@3.38.5%gcc@11.3.1+column_metadata+dynamic_extensions+fts~functions+rtree arch=linux-fedora35-skylake
^util-linux-uuid@2.37.4%gcc@11.3.1 arch=linux-fedora35-skylake
^py-setuptools@59.4.0%gcc@11.3.1 arch=linux-fedora35-skylake
^py-wheel@0.37.0%gcc@11.3.1 arch=linux-fedora35-skylake
Information on your system
- Spack: 0.18.0.dev0 (d7d0c89)
- Python: 3.10.4
- Platform: linux-fedora35-skylake
- Concretizer: clingo
General information
Steps to reproduce
When building a new package which depends on
py-numbaspack concretizes to use the oldest available versionpy-numba@0.35.0instead the newest onepy-numba@0.55.1without any reason. Therefore also downgrading python to version 3.6.15Reproduce:
~/.spack)py-numbadependencyI used
py-fracridge(the package I wanted to create when encountering this problem) with thepackage.pyping @adamjstewart @alalazo
Error message
Information on your system
General information
spack debug reportand reported the version of Spack/Python/Platform