Fix matplotlib and update to v3.1.3#2076
Conversation
|
|
||
| Because, for unix platforms, the mpl install script uses `pkg-config` | ||
| to detect libraries installed in non standard locations (our case... | ||
| well...we don't event install the libraries...so we must trick a little |
There was a problem hiding this comment.
we don't event -> we don't even
| def get_install_requires(self): | ||
| - return ['numpy>=1.10.0'] | ||
| - return ['numpy>=1.11'] | ||
| + return ['numpy==1.16.4'] # to match p4a's target version |
There was a problem hiding this comment.
I have some doubts with this line. Does it really have to be hard pinned to the current default p4a numpy version?
And if it does, should we at some point try to pick this up dynamically?
There was a problem hiding this comment.
I have some doubts with this line. Does it really have to be hard pinned to the current default p4a numpy version?
I'm not sure, I'm doing a local test...we will see
And if it does, should we at some point try to pick this up dynamically?
I think so, but it will add more complexity to the recipe...
There was a problem hiding this comment.
Yes I was also thinking about the complexity. I think we can address that later if needed.
You have my ✔️ feel free to merge if it builds 💚
AndreMiras
left a comment
AndreMiras
left a comment
There was a problem hiding this comment.
LGTM, the Python 2 armeabi-v7a (with numpy) build failed on network IO, so I restarted
Because the `matplotlib` recipe should work now in our `CI` tests
|
Thanks for the Note: I updated the first post, since with the changes made in here we can close some issues. |
|
I think that this is ready to 🔀 let's do it!! 😸 |
2 participants
This PR depends on a working
kiwisolverandcppyrecipes (#2075).Closes: #1900, kivy/buildozer#1003
Notes: I will merge the branch of this PR alongside the branch of #2075 into #2073 so we can see if the thing is working with the
on_device_unit_test_app.See also: