[RELEASE] 0.18.2 bump version, added whatsnew #9167
Conversation
|
looks like docs are not pushed because stock-market is still failing :-/ |
|
I guess the urlopen is not a context on python2? |
|
plot_stock_market is breaking Circle CI at master too
https://circleci.com/gh/scikit-learn/scikit-learn/11448
…On 20 June 2017 at 09:40, Andreas Mueller ***@***.***> wrote:
I guess the urlopen is not a context on python2?
—
You are receiving this because you are subscribed to this thread.
Reply to this email directly, view it on GitHub
<#9167 (comment)>,
or mute the thread
<https://github.com/notifications/unsubscribe-auth/AAEz6zfKUYFrs8oR9wXS1KIZXrUbN-dZks5sFwb3gaJpZM4N-v8O>
.
|
|
ok there seems to be another unrelated failure... |
|
sphinx-doc/sphinx#3212 what? |
|
@jakirkham I spent about 3 days on trying to make this release happen and now I'm going away for a week, not sure if anyone will pick it up.... |
|
Ouch! Sorry to hear that. On the one hand I'm curious to know what still needs to be done to make this patch release possible as getting scikit-learn built with NumPy 1.13 would be great. However I don't know how helpful I will be as I know much less about the goings on here. On the other hand, assuming things go as planned, there will be a new release of scikit-learn in a couple weeks with NumPy 1.13 support IIUC. Also we had talked about doing an RC or so at conda-forge this go around. Given all of this, maybe we should just punt on the issue and just start preparing for scikit-learn 0.19.0? What do you think? |
|
I'll try for the next ~1.5 hours that I'll have this week, and then I'll leave it up for the others to decide. |
|
@amueller I am trying to see if I can go somewhere for the next hour with the ci |
|
@glemaitre thanks :) |
|
Hurray. |
|
Thanks @amueller! |
|
@jakirkham I pushed a tag, the wheels are building. So feel free to use the tagged version for conda forge, I think? |
|
Great, thanks. Have opened PRs at conda-forge to update scikit-learn based on the tag. Will try and get NumPy 1.13, 1.12, and 1.11 builds of 0.18.2 out in the next day or so. |
|
thanks :) |
Mostly to run the integration ;)
This is the tiniest release ever, mostly to ensure conda forge doesn't need to patch and still works on current numpy.