validate numpy1.12 compatibility changes #5527
Conversation
|
jenkins test this |
|
Splitting off |
|
Seem ok after the revert of the version changes and disabling of blas check.
…On Fri, Feb 10, 2017 at 2:09 PM Pascal Lamblin ***@***.***> wrote:
Splitting off test_abstract_conv.py in a new PART seems to have worked
for Travis, and the tests pass.
I'm going to revert the changes to the numpy and scipy version so it can
be merged.
@nouiz <https://github.com/nouiz> or @slefrancois
<https://github.com/slefrancois> could you please have a look at my
commits?
—
You are receiving this because you were mentioned.
Reply to this email directly, view it on GitHub
<#5527 (comment)>, or mute
the thread
<https://github.com/notifications/unsubscribe-auth/AALC-6ZjewgZI4XEF5a9VP_r3vlinNYxks5rbLXRgaJpZM4L9Aeg>
.
|
Fix CAReduce.c_code_cache_version_apply Calling `append` on a tuple raised AttributeError, which was mistaken for a missing `c_code...` function altogether, and the version was ignored.
casting to int could lead to issues in the following case: - 64-bit Windows - python 2 - shape bigger than 2**31 In that case, int would be 32-bit, but shapes could be bigger and stored in long.
lamblin
force-pushed
the
ccw_numpy_travis
branch
from
a89e476
to
4631b6e
Compare
|
I think I finally got down to the cache issue. |
lamblin
force-pushed
the
ccw_numpy_travis
branch
2 times, most recently
from
1d93772
to
eca5cab
Compare
|
So everything works fine, as long as the non-conda packages for scipy and numpy are not installed in Travis' cache (in which case it makes linking with blas fail), which I did by temporarily removing I'm confident everything is fine now, so I'm restoring |
|
Actually, Jenkins was complaining as well. |
lamblin
force-pushed
the
ccw_numpy_travis
branch
from
eca5cab
to
848fae5
Compare
|
And this last one should fix the version for |
This is just a rebase of #5461 to see if the issue happens with the main cache.