kammerje / fouriever Goto Github PK

There is a lot of examples data in the repo, which is great! However, it makes cloning or installing Fouriever directly from github with pip very slow. I was wondering if there were other option available to store some of the data, maybe keeping one or two examples in the repo for quick tests, but making the bulk of the data available on another platform (OneDrive, Box, NextCloud, Dropbox, etc., or even a seaparate git repo, for example fouriever-data).

I could contribute with the OneDrive storage I have at UdeM or by settings up another repo if you want @kammerje.

Hi @kammerje,

With the upcoming KPI pipeline for JWST and data starting to be available, it might be good to have an official release on PyPI to make it easier to:

1. Find fourieverand install it
2. Include fouriever as a dependency to other projects

Happy to help with this if needed.

Currently, figures are saved and closed internally. It would be nice to have the option to show the figure as well (with a switch in the function argument, or to make the figure accessible for users, so they can show it with plt.show(). This would especially nice for notebooks (e.g. the KPI Pipeline example).

Hi,

I encountered a few edge-cases when saving files I thought I'd combine in one issue:

1. Some directories (e.g. caldir) are assumed to end with a separator (/) when combining them with file names. Depending on how users create the path (e.g. os.path function or Path objects), this could cause unexpected behaviour like /path/to/my/directorywithfile.ext.
2. Related to the above, using str.rfind("/") to find directory can sometimes be error-prone if users pass paths from os.path function or Path objects or work with Windows (sep=\)
3. This might be a personal preference, but I file like the notation ofile for directory + file prefix can be confusing for users. I would have intuitively expected it to be the full file path or name (including extension).

Regarding the last point above, maybe separating ofile in something like odir and file_id that makes it clear what part is a (directory) path and what part is an ID used to generate file names would clarify things for new users? A potential way to implement would be to keep ofile and add new options for a future release, with a deprecation warning, and then fully replace a few releases after.

@kammerje let me know what you think of the above. Happy to send a PR implementing some or all of these suggestions.

Thank you!

The chi2map_sub method will give the following error when chi2map was not executed beforehand, even though the return from chi2map was stored and used as input for chi2map_sub:

KeyError: 'uu_smear'

Perhaps because chi2map_sub requires an attribute that was created in chi2map?