spec: fix ABNF grammar and add label normalization SHOULD#148
spec: fix ABNF grammar and add label normalization SHOULD#1480xheartcode wants to merge 2 commits intoremenoscodes:mainfrom
Conversation
Section 4.8 defined CRLF = %x0A which contradicts RFC 5234. Renamed to LF throughout the grammar. Also tightened title to 1*TEXT-NO-LF to reject empty titles and updated the 72-char guidance to SHOULD. Closes remenoscodes#130
Implementations SHOULD normalize labels to lowercase before comparison to avoid case-variant duplicates. Not REQUIRED, to preserve intentional casing like iOS, API, UI. Referenced in Section 6.3.1. Closes remenoscodes#133
remenoscodes
left a comment
remenoscodes
left a comment
There was a problem hiding this comment.
The ABNF fix (CRLF → LF) and the 1*TEXT-NO-LF tightening look correct — thank you for catching the RFC 5234 contradiction.
On the label normalization SHOULD: there's a tension with the existing normative text. Section 4.7 currently states "Labels are case-sensitive: Bug and bug are distinct labels" and the merge comparison says MUST compare "exactly (case-sensitive)." Adding a SHOULD to normalize to lowercase creates conflicting guidance — an implementor reading both statements can't satisfy them simultaneously.
Could you clarify the intended interaction? Options:
- Normalization applies at creation/edit time (deduplication), while merge comparison remains exact
- The case-sensitive statement gets softened to match
- Split into a separate PR so the uncontroversial ABNF fix can land first
Option 3 would also let us merge the ABNF fix immediately.
remenoscodes
left a comment
remenoscodes
left a comment
There was a problem hiding this comment.
Following up on the earlier review — the ABNF fix (CRLF → LF, 1*TEXT-NO-LF) is solid and I'd like to merge it soon.
To unblock that, could you split this into two PRs?
- PR A: ABNF grammar fix only (LF rename + title tightening) — ready to merge immediately
- PR B: Label normalization SHOULD — needs the normative conflict resolved first (see previous comment about the tension with the existing case-sensitive MUST)
This way the uncontroversial fix lands now, and we can iterate on the label normalization separately without blocking it.
Also, both commits should follow conventional commits format: spec: fix ABNF grammar to use LF token and spec: add RECOMMENDED label normalization.
2 participants
Two spec-only clarifications to
ISSUE-FORMAT.md, no implementation changes required.Changes
Fix ABNF grammar to use LF token (closes #130)
Section 4.8 defined
CRLF = %x0Awhich contradicts RFC 5234 (where CRLF means%x0D.0A). Renamed the token toLFthroughout the grammar to accurately reflect that Git commit messages use LF line endings. Also tightenedtitlefromTEXT-NO-LFto1*TEXT-NO-LFto explicitly reject empty titles, and updated the 72-character guidance to useSHOULDper RFC 2119 convention.Add RECOMMENDED label normalization (closes #133)
Added a
SHOULDstatement to Section 4.7 that implementations normalize labels to lowercase before comparison, to avoid case-variant duplicates likebugvsBug. Marked as RECOMMENDED not REQUIRED to preserve intentional casing (e.g.,iOS,API,UI). Referenced from Section 6.3.1 where the case-variant limitation is documented.Test plan
make test: 282 passed, 0 failed