Skip to content

Latest commit

 

History

History
78 lines (45 loc) · 2.73 KB

2022-05-06.md

File metadata and controls

78 lines (45 loc) · 2.73 KB

SPDX Canonicalisation Committee Meeting - May 6, 2022

Attendees

  • Alexios Zavras
  • David Kemp
  • Henk Birkholz
  • Joshua Marpet
  • Maximilian Huber
  • Nisha Kumar
  • Saul Wold
  • Sebastian Crane

Agenda

  • Minutes approval; discussion of new process
  • DateTime property
  • AOB

Notes

New minutes approval process

Sebastian will upload minutes to GitHub shortly after each meeting and ping attendees for review. If there are no outstanding objections at the next meeting, Sebastian will merge the GitHub pull request. This is the same process as used by the Defects Team.

DateTime property

DateTime is used in every element. Options include ISO timestamp (UTC), or integer of seconds/milliseconds since Unix epoch.

  • Alexios: would advise against ISO - many representations. RFC 3339 is a more specific version of the ISO format.

  • David: would prefer integer since Unix epoch, seconds or milliseconds although not sure what the milliseconds' use case is.

  • Maximilian points out that an unsigned integer can not be used for times before Unix epoch.

All agreed that there might be used cases for digital artefacts that were created before epoch.

  • Joshua asks which option is easier to adopt.

  • Sebastian: one issue is space-time calendar for space exploration where the Gregorian calendar is not used

  • Maximilian: is the criteria only for a machine readable format or can it be human readable?

  • Henk: human representation is disconnected from the machine readable - you can convert it to human readable

  • Nisha: since the use case is canonicalisation, would prefer epoch.

Preference of those present is for a signed integer of milliseconds since Unix epoch.

  • Nisha: would suggest choosing one JSON standard as the implementations vary

  • Henk proposed IETF Standard 90, which has a grammar that is widely accepted.

  • Options considered:

  • Max: would like to target the minimum that satisfies many other specifications

  • All agreed to use IETF Standard 90.

Decisions

  • We will use IETF Standard 90 with additional restrictions if necessary.

  • If we come up against problems that I-JSON solves we can change what we've written to I-JSON.

  • We'll define that objects that never have duplicate keys in the canonical serialisation.

Action points

Topics for next meeting

  • SemVer property