Some Eclipse Foundation services are deprecated, or will be soon. Please ensure you've read this important communication.

Bug 346016

Summary:	EOL Editor Supported Comments
Product:	[Modeling] Epsilon	Reporter:	Horacio Hoyos <arcanefoam>
Component:	Core	Assignee:	Antonio Garcia-Dominguez <agarcdomi>
Status:	CLOSED FIXED	QA Contact:
Severity:	minor
Priority:	P3	CC:	agarcdomi, dkolovos
Version:	unspecified
Target Milestone:	---
Hardware:	PC
OS:	Windows 7
Whiteboard:	interim

Description Horacio Hoyos

2011-05-16 20:04:01 EDT

When editing epsilon languages the editor fails to correctly recognize the end of multiline comments when using the /* ... */ or the -* ... *- style. This happens either with new comments or when editing existing ones. If the file is closed and reopened the editor recognizes the comment ends correctly. 

The bug is detected because after the comment is closed the rest of the text in the file remains colored as comments.

Comment 1 Antonio Garcia-Dominguez

2011-05-17 04:52:31 EDT

Thanks for your report! I'll have a look today at the bug.

Comment 2 Antonio Garcia-Dominguez

2011-05-17 07:39:10 EDT

This is a follow up to this thread in the forum:

http://www.eclipse.org/forums/index.php/t/209207/

I think I have fixed the problem in SVN r1422. The fix itself is trivial, but finding the part that required fixing took a while.

The partition scanner we use to divide the program into comment and non-comment sections has two JFace MultiLineRules for the -* *- and /* */ comments. Normally, these rules look for their start delimiter and then for their end delimiter. In this case, the rules do not conflict, as they start with different characters.

However, there's one case in which PatternRule#doEvaluate skips looking for the start sequence: when "resuming" the partition scanning. This happens when we insert a newline inside a comment, for example. PatternRule sees the * in *- and mistakes it for the * in */. Since breakOnEOF is set to true for that rule, it plows through the rest of the file and accepts EOF as an end sequence.

I have fixed this bug by setting breakOnEOF to false, so it will properly reject the */ rule and go to the *- rule.

Comment 3 Dimitris Kolovos

2011-07-25 08:18:04 EDT

Fixed in 0.9.1