A comment on a review is currently uniquely identified by:
- The SHA of the commit
- The path of the file relative to this SHA
- The line number relative to the file
The diff hunk that is also found is redundant and could be re-constructed from the SHA and its parent. The rest is meta information about the comment (author, creation date, update date, reactions).
This can only represent a comment that applies on a single line. Not to a range of lines. The number of lines should be added for that purpose. If it cannot be represented in a forge, it can be ignored. Some information is lost but the comment itself still can be inserted where it has meaning. For instance Forgejo only knows to represent review comments on a single line but GitHub review comments can be related to a range with the line and start_line data fields.
modified reviewcomment.json
@@ -25,6 +25,10 @@
"description": "The line number of the comment relative to the tree_path.",
"type": "number"
},
+ "lines_count": {
+ "description": "The range of lines that are commented on. If absent it defaults to one and is a single line comment. If specified it must be a positive number. If line is N and lines_count is C, the range of lines commented on is ]N-C,N]. In other words, the range starts lines_count before line, which is the last of the range",
+ "type": "number"
+ },
"commit_id": {
"description": "The SHA of the tree_path commented on.",
"type": "string"
