2025-12-18 meeting

[rlskoeser]

Agenda

• Hear from Jason Heppler about upcoming projects that could benefit from undate
  • Discuss requirements / use cases for dates in these projects (storage, parsing, import/export, search, filtering, calculations)
• Discuss & decide approaches for database implementation
  • brief overview of past approaches from other projects
• Update from @rlskoeser on recent work & possible next steps
  • Identify any priorities
• If time: discuss what is needed to get to a 1.0 / sufficient maturity to submit to pyOpenSci

Notes

attending: Jason Heppler, Robert Casties, Julia Damerow, Rebecca Koeser

use cases for Jason's projects (past/future/possible)

• Mapping Violence early modernist working on violence in Italy; court records, cases, bureacratic cases - complicated; partial dates, not always well-recorded; not different calendars in this case. (earliest 1520, runs about a century); code : https://github.com/chnm/mv-django
  • variability in precision, but still match on or filter dates, query for related cases by date
  • django requires precision in dates
    • two fields: what's on the document; then python-edtf for the dates - but it's clunky and confusing to work with
      • separate django app for historical dates — https://github.com/chnm/mv-django/tree/main/historical_dates
        • display date (date on document); fuzzy date (text field, optional), start date and end date (range if you don't know the month); drop-down for certainty (exact, approximate, uncertain); note field
• London plague bills - may be moving data into django; some of the dates are fuzzy/incomplete/unclear;
  • during this time, the calendar changed - have a split year, so record as 1665/1666; lining that up has been tricky; would be great to convert or pin into a calendar (Julian / Gregorian calendar)
  • not a a Django project but you can see some of how dates are handled here: https://deathbynumbers.org/database/?start-year=1629&end-year=1754&bill-type=weekly&primary-tab=annual&secondary-tab=parishes&tab=1
    • how complicated - can you assume people switched at the same time?
      • documents produced by the monarchy, can probably assume standard - but other projects might have
• Connecting Threads - textiles in the global south (code: https://github.com/chnm/ct-django)
  • track when something was made ; movement of items across time and space
  • varied precision of dates, wider range of dates
• bibliographic data: full date then text field (more flexible precision); sorting and filtering on isodate (full precision); add tags for century
  • needs: sorting, searching, displaying by year
  • all in frontend and model, no django admin - but if it's easy to switch would be an easier sell

other / prior implementations

• Shakespeare & Company Project - store as date with flags about which parts are known; django custom fields and filters for querying on known year; also custom template tags for rendering dates without losing ambiguity
• Princeton Geniza Project - converting from Hebrew, Seleucid, Anno Mundi to standard; display in common era (Julian/Gregorian); stores original date, then parses to convert and store as EDTF (date or range);
• ISMI - important to preserve original date and also the uncertainty; RDF database - CIDOC-CRM

best implementation?

• text string with EDTF seems the most flexible - could start with that and try to build django utilities around it
• set of documents with partial information:
  • recognize as a date, run queries on, display as a date
  • filtering, searching, expose the uncertainty
    • dataviz that exposes the uncertainty
• easy for users to enter and see as a date (human readable)
• flexibility to sort unknowns / first last
• how to preserve knowledge about one point in time vs uncertain range

structural: keep within undate package as an optional dependency; test with and without django

• would be nice to have some examples of how to apply it
  • e.g., could be multi-field like geniza implementation, but not all those fields are required
• support multiple different ways of combining the fields, displaying the fields

Next steps

• Jason & Rebecca will schedule working sessions to plan and start building out django support in undate

[rlskoeser]

Data Downloads from the Death by Numbers project could be useful to experiment / check against undate:

• https://deathbynumbers.org/downloads/