2024-10-03 co-working session

[rlskoeser]

Instead of a meeting with questions and discussion, we will use our regular monthly meeting as a co-working session to review open pull requests and work on smaller tasks that we can complete or nearly complete during the meeting.

Pull requests awaiting review

• preliminary str parse and format methods #90
• Make EDTF parser available as undate formatter; handle 5+ digit years #89
• Use numpy datetime64 instead of datetime.date #77 #84

Issues that could be worked on

Any open and unassigned issue in the 1.0 release is up for grabs.

low hanging fruit

Some possibilities that we may be able to finish or at least get a good start on in a short time:

• Write up branching guidelines in contributing.md #55
• document how to add new formatters #86
• EDTF support - level 0, 1 and parts of level 2 #73
  • start writing a Jupyter notebook that demonstrates using EDTF parsing and formatting as a way to start checking off the parts of the spec we are compliant with.
• measure and document how the numpy dependency changes the package - currently tracked on integrate numpy datetime64 for greater range of years #77 but we could make a separate issue

stretch possibilities (high-hanging fruit?)

More complicated/interesting issues that could be valuable to start on together but we're unlikely to finish:

• implemented converted date for dates from other calendars #10
• support fuzzy/approximate dates with configurable fuzziness  #8
• handle dates with missing information, e.g. due to corrupted or missing text  #64

[rlskoeser]

@jdamerow and I added branching documentation to contributor docs to address #55 . We merged to develop but did not close since the feature is not released. Added a new issue label "resolved" to track this status. Since the issue was referenced in the PR and commit message, it should automatically close when the changes are merged to main.

We should think about where to document this convention, unclear if it belongs in contributing docs or elsewhere.

[rlskoeser]

Discussion for future work: parsing human readable dates sounds like a task for an LLM. Would be valuable to experiment with this as an optional feature for undate, maybe few shot learning with an api key, maybe an open source model could work.

Also interesting for multilingual.

Collect 10-20 examples and experiment in a notebook to start.