A few unrelated bugfixes #15
Conversation
|
I will let @schlafly do the review, but there is some problem associated with versioning in conda that has caused issues with the conda-forge feedstock for crowdsource that is causing my CloudCovErr.jl unit tests to fail, so I releasing a new crowdsource package was on my to do list anyway. Happy to handle that part of it. |
|
Thanks for the PR. Let me review and merge tomorrow. I may tweak the get_sizes(...) implementation, but it's good to have that configurable now. I don't understand python packaging well enough; Andrew, can you confirm the change to the name in the setup script? Thanks! |
|
Can you see if this API for get_sizes(...) fits your needs? I also took the liberty of adding some line breaks. I haven't been being careful about print(..., flush=True); if you're having issues there, you might consider starting python in unbuffered mode (python -u), rather than adding to the print messages. |
|
the I think the API works but I'm not sure I know what In any case, yeah, I think this adds the flexibility needed. |
|
Yes, in DECaPS / WISE we occasionally observe alpha Cen / whatever and it's possible for hundreds of pixels to be masked. We try to do a little better in those cases by using the full 299x299 stamp we have. That's only new in that I moved the hard coded constant into a default parameter. |
|
@schlafly Looks good to me. I built and tested on TestPyPI and had no issues. I just bumped the patch version number. After you merge, I will issue a Github release, push to PyPi production, after which it should appear on conda-forge same day. My real concern is that I made CloudCovErr.jl depend on crowdsource (a bad idea in retrospect) and I am worried about downstream compatibility issues. So I am fine with going ahead with this as long as you are ok with some possible patch version number incrementing while I troubleshoot that in the next few days. |
|
Great, merging. |
If you want, I can split these up. The most substantive changes are re: #14. Others are things I put in to help me track the progress of the code; these could be removed. There's also an import change (that's a real bugfix) and a change to the packagename in the setup.