diff options
| author | Egor Tensin <Egor.Tensin@gmail.com> | 2020-11-29 21:10:29 +0300 |
|---|---|---|
| committer | Egor Tensin <Egor.Tensin@gmail.com> | 2020-11-29 21:10:29 +0300 |
| commit | cdf542502879ef992efa59dce788e703dd978578 (patch) | |
| tree | 14b2833b5aa5fc5c51bfe63bb823edf81a68db0c | |
| parent | fix 3.8+ errors (diff) | |
| download | vk-scripts-cdf542502879ef992efa59dce788e703dd978578.tar.gz vk-scripts-cdf542502879ef992efa59dce788e703dd978578.zip | |
clarify dependency version bounds
| -rw-r--r-- | constraints.txt | 10 | ||||
| -rw-r--r-- | requirements.txt | 1 |
2 files changed, 9 insertions, 2 deletions
diff --git a/constraints.txt b/constraints.txt index 8a74c51..04e6c11 100644 --- a/constraints.txt +++ b/constraints.txt @@ -1,3 +1,9 @@ -# Numpy 1.17 dropped support for Python 3.4. -numpy >1.11,<1.17 ; python_version == '3.4' +# matplotlib 1.5 is one of the versions that works with Python 3.4, and it +# doesn't specify the upper bound for its numpy dependency. So pip will try to +# install the latest version, which doesn't work with 3.4. Putting the numpy +# and the matplotlib requirements to the same requirements.txt file doesn't +# work unfortunately, the latest version will still get picked up. + +# The lower bound is random, 1.16 is the latest version to support 3.4. +numpy >=1.11,<1.17 ; python_version == '3.4' numpy >=1.17 ; python_version > '3.4' diff --git a/requirements.txt b/requirements.txt index 0e97461..9e942a5 100644 --- a/requirements.txt +++ b/requirements.txt @@ -1,3 +1,4 @@ +# Some random known-to-work versions. matplotlib ~=1.5 ; python_version == '3.4' matplotlib ~=2.1 ; python_version == '3.5' matplotlib ~=3.1 ; python_version > '3.5' |