New issue
Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.
By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.
Already on GitHub? Sign in to your account
Upgrade to NumPy 1.15.2 #25260
Comments
comment:1
For NumPy 1.13.3 we have the tarball in .zip, should we also use the .zip this time or the .tar.gz? |
This comment has been minimized.
This comment has been minimized.
comment:2
We now support zip files I believe, so this is fine. |
comment:3
If we support both kinds of archive, we should probably use the smallest one. |
comment:4
I don't know how or if I can add changes here, but I have fixed the doctests for the numpy+scipy upgrade here. Its mostly formatting. Interesting bits:
Edit: I was a bit too fast and didn't test it properly. I'll post an updated version once thats tested. |
comment:5
I have updated the patch. |
comment:6
Changing target from Numpy 1.14.3 to 1.14.4. |
This comment has been minimized.
This comment has been minimized.
Commit: |
Branch: public/25260 |
comment:7
I updated the package and applied my doctest fixes. New commits:
|
This comment has been minimized.
This comment has been minimized.
comment:8
Numpy 1.15.0 just came out. |
Dependencies: #25755 |
comment:12
I've done most of the upgrade, but two failures in
This does not happen outside of sage. It probably has something to do with sages custom number types. |
Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:
|
Branch pushed to git repo; I updated commit sha1. This was a forced push. New commits:
|
comment:69
Just for the record I would prefer to monkey-patch instead (or maybe additionally). It seems to be a clear bug in this case and downstream distributions should probably patch, but I still think sage should do what it can to prevent errors even on an unpatched system. I know your views on this but still wanted to make that point. |
comment:70
gh-timokau, I would usually agree that we should monkey-patch. However, in this case, there seems to be nothing Sage specific going on so I do not see a strong reason to monkey-patch. In particular, this does not completely break Sage but just breaks things in some rare cases. As this is a numpy bug, distributions should backport it in their numpy if they care. Probably more importantly, this would be somewhat hard to monkey-patch. Numpy is not imported in a central spot in Sage as the import takes a while, so it is only loaded on demand. Therefore, there is not a single spot where the monkey-patching could happen. |
comment:71
Yes, I agree that the argument is a bit weak in this case. If it was easy I would still prefer it, but since its not its not that bad. |
Upstream: Fixed upstream, but not in a stable release. |
comment:72
The patch was accepted upstream. |
This comment has been minimized.
This comment has been minimized.
comment:73
Should we then upgrade to NumPy 1.14.6 now, or wait for NumPy 1.15.3 to be released? |
comment:74
From a packaging perspective its preferable to upgrade to the latest version now. The latest version just requires a patch for some edge case. Without the upgrade sage is completely incompatible with newer numpy versions (which distros probably have by now). Thank you for your efforts in fixing this upstream Jeroen! |
Changed branch from public/25260 to |
comment:76
This should be re-targeted for 8.5. |
Changed commit from |
comment:77
Next NumPy upgrade at #26643. |
comment:78
I get two doctest failures on OSX with SageMath version 8.5.beta2, which I believe are related to this ticket, please let me know what other information is helpful (or if I should ignore this).
|
comment:79
I don't think it is the first time we have a formatting issue of that kind but I am not sure how to deal with it. The first thing that comes to mind is: are you using OS X' clang or sage's gcc? |
comment:80
I'm not 100% sure I'm looking in the right place but looks like homebrew gcc, version 8.2.0. As far as I know I haven't tried to change to clang for my sage install. I have this line in config.log |
comment:81
Hum... Not sure. You should open a new ticket in any case. While numpy could be your culprit if numpy arrays are used it could be something else entirely. |
NumPy 1.15.2 was released.
Tarball: https://files.pythonhosted.org/packages/45/ba/2a781ebbb0cd7962cc1d12a6b65bd4eff57ffda449fdbbae4726dc05fbc3/numpy-1.15.2.zip
Our last upgrade was to NumPy 1.13.3 in #24063.
A bug introduced in NumPy 1.15 prevents building NumPy on machines with many CPUs.
The issue is reported and solved upstream:
Upstream: Fixed upstream, but not in a stable release.
CC: @kiwifb @infinity0 @jdemeyer @slel @sagetrac-epalezzato
Component: packages: standard
Keywords: upgrade, numpy
Author: Timo Kaufmann, Julian Rüth, Bryan Gin-ge Chen
Branch:
b151634
Reviewer: Travis Scrimshaw
Issue created by migration from https://trac.sagemath.org/ticket/25260
The text was updated successfully, but these errors were encountered: