Clarify symbol boundary rules for digits#370
Conversation
Add spec text clarifying that colons directly adjacent to digits on both sides do not open or close symbol syntax. This prevents time formats like 10:30:00 from being misinterpreted as containing symbols (e.g., :30:). Addresses jgm#350.
|
#350 also discussed (in comments) IPv6 addresses (say |
|
You're right that this is a targeted fix for time formats specifically, not a complete solution for all colon-containing strings. IPv6 addresses and URNs are valid edge cases that this doesn't cover. The digit-boundary rule was chosen because time formats (like 10:30:00) are by far the most common case where users unexpectedly hit symbol parsing. IPv6 and URNs are less common in prose and can be handled with escaping or raw/verbatim spans when needed. A more comprehensive solution would risk breaking legitimate symbol uses. I've updated the PR description to clarify this addresses the time format case specifically. |
Summary
10:30:00from being misinterpreted as containing symbols (e.g.,:30:)Partially addresses #350.
A reference implementation already exists in djot-php, closing the gap between upstream spec and downstream implementations.