Hello Pierre,

I can suggest implementing different matching modes for the FoldingTaggerBase class. The default mode will work like the current version, assuming the document folding regions are valid - equal number of start and end tags, no crossings between regions. This mode now runs in linear time which is fast (stack iteration). However, to match tags in non-valid document, code will be a bit slower (but will cover cases like yours), because of the need to check all possible end tags (and they are not ordered). So there can be at least 2 modes - 'first' and 'last' match. I am attaching here your modified project with this new API which is in development. So it can be changed over the next weeks. I would appreciate any feedback.

    var foldingTagger = new FlexFoldingTagger(this.syntaxEditor) { MatchingAlgorithm = FoldingMatchingAlgorithm.FirstMatch };

Regards,
Petar Mladenov
Progress Telerik

Hi Pierre,

To be honest this is logged as a bug but is more like a not supported feature currently. For example in some well-known languages error will be shown in the IDE /Compiler if the structure of the language does not follow the rules. Our SyntaxEditor misses such error / (syntax) validation feature Currently I have a fix candidate code but I am not sure all clients will expect the behavior. For example: Open -close tag couples are A1-A2 and B1-B2. In the following structure:

A1
    B1
A2
A2

what do you expect - folding from A1 to first A2 or folding from A1 to second A2. I would appreciate your feedback on the matter. Also, probably property will be a good addition to the FoldingTaggerBase class to cover both scenarios. Having your input in mind, I will raise a discussion in my team and will let you know how we will proceed with this item. Thank you for your understanding.

Regards,
Petar Mladenov
Progress Telerik