Unexpected highlighting behavior when end is not defined

[aeschli]

From @mhchem on April 25, 2018 9:20

With this simplified .tmLanguage file, the whole document should be formatted a comment (because I use a begin without and end).

<?xml version="1.0" encoding="UTF-8"?>
<!DOCTYPE plist PUBLIC "-//Apple//DTD PLIST 1.0//EN" "http://www.apple.com/DTDs/PropertyList-1.0.dtd">
<plist version="1.0">
    <dict>
        <key>scopeName</key>
        <string>text.test</string>

        <key>patterns</key>
        <array>
            <dict>
                <key>begin</key><string>^.</string>
                <key>name</key><string>comment</string>
            </dict>
            
            <dict>
                <key>begin</key><string>.</string>
                <key>name</key><string>invalid</string>
            </dict>
        </array>
    </dict>
</plist>

This works fine unless the document contains the word undefined.

I guess this is a "end is undefined" issue where undefined is taken literally.

The issue can be avoided by setting end to a never-matching regexp (like <key>end</key><string>^\b$</string>). And maybe that could be put into the highlighting code: if undefined, use ^\b$ instead of the string undefined.

Copied from original issue: microsoft/vscode#48654

[alexdima]

I think we never ran into this because a grammar with a begin statement and a missing end statement is illegal... So our resolution could be to throw the entire rule out.

[mhchem]

It's your decision. All I can say is that I read the documentation differently. (My understanding of "If there is no match for the end pattern, the end of the document is used." was that I can leave out the end to achieve a match till the end of the document.)