In authoring #75 I noticed some potentially annoying edge cases we don't handle correctly:
// FIXME: this is current behaviour, but we should probably error on
// out-of order durations (not RFC3339 compliant)
duration_all_wrong_order: ok => "P3D2M1YT6S5M4H", "P1Y63DT4H5M6S";
// FIXME: this is current behaviour, but we should probably error on
// repeated units (not RFC3339 compliant)
duration_unit_repeated: ok => "P1Y2Y", "P3Y";
We should decide if we want to clean up our parsing to not accept these cases, and if so, how we deal with the fact this will make parsing stricter for users.
In authoring #75 I noticed some potentially annoying edge cases we don't handle correctly:
We should decide if we want to clean up our parsing to not accept these cases, and if so, how we deal with the fact this will make parsing stricter for users.