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
Deprecation warnings and test failures under Python 3.8 #8938
Conversation
The I'll defer to @nden for the |
@bsipocz , does Python 3.8 stuff get backported? How far back? EDIT: Nvm... Saw your milestone. LOL |
I'm on the side to backport these kind of fixes while they are small an painless, and draw the line when we run into some uncomfortable situations. I don't think we should aim for LTS, but 3.2 may still make some sense. |
Forgot that we still support Python 3.5 (or rather confused it with matplotlib 3.1) - tested with 3.5.7 under OS X 10.12 as well now. I could not reproduce any of the failures on circleci 32bit, but note that my local tests were against this PR + @astrofrog's |
There was a problem hiding this comment.
Choose a reason for hiding this comment
The reason will be displayed to describe this comment to others. Learn more.
The wcs
change looks fine too. velref
is defined as integer.
This doesn't exactly have the same failures as on master for 32bit, so I would hold up merging until at least master is sorted out. |
How can one access the CI output for |
on circleCI you can select the master branch on the left hand side, on travis select the "build history" tab. |
Ah, thanks, found it (in my browser it's on top). |
@dhomeier - please rebase |
Tried to exactly follow the manual in rebasing this time; still looks messy, sorry! |
Hmm, as I see you've rebased it on the master of your fork rather than on astropy master. This is basically the main reason why we suggest to remove the master branch of the fork :) This is how I would do it:
make sure you're on the branch:
This is a clear case, so the interactivity is not even needed, but nevertheless it's good to get into the habit of using the interactive way. So, now you have an editor window with the 2 commits listed. Save and exit it. And finally the force push: |
Hmm,
you mean delete the master branch altogether in the local checkout (and
was exactly equivalent (even implicitly checking out the topical branch)... And I |
4ff4126
to
912ed85
Compare
912ed85
to
104fd72
Compare
Mmh, I also did not wish to rewrite history, but this seemed the best way out now, as I did not want to decide on merging others' commits either. Still don't quite get the stupid warning about
as there are no remote changes I wish to add locally (OK, I do see that the remote tip was |
the thing that we you did that from your origin/master it wasn't the actual top of the master branch. |
you can't push to |
A rebase rewrites history, so in almost all cases you need to use |
Yep, the |
|
Thanks @dhomeier! |
Thanks @bsipocz, never mind my ranting, I was just keeping notes of what I found odd or confusing in the work pattern (and where I would have git expected to be smarter)! |
@dhomeier - PRs are welcome to our dev documentation. It's actually helpful if when people are pointing out which part was confusing or not helpful enough. |
Deprecation warnings and test failures under Python 3.8
Two more minor fixes for DeprecationWarnings appearing with 3.8b1:
First is the
from the
votable.tablewriter
call inio/votable/tree.py
, causing most votable tests to fail.Fixed by defining PY_SSIZE_T_CLEAN.
The second is a
as
_wcs.Wcsprm.velref
is an int, trivially fixed by setting it with an integer (raises the question whether the results should be compared to integers as well) - unless we wishvelref
to accept floats, of course (or rather want to filter the warning, as the removal notice is still vague).Tested for backwards compatibility with Python 3.6.