Indeed it was, and the fix no longer applies. I'm thinking aloud to how I could get things working again.
Basically, the idea is this: at line 70, one computes the value for startPos so that the string between there and curPos is looked up in the API file. It is computed using whatever Scintilla currently considers as part of a word. Scintilla needs to be told some extra characters are word like as well.
Current Scintilla can be told so by sending it SCI_SETWORDCHARS with the unusal 63 char long string concatenated with the additionalWordChar string, just line in NppNotifications:756 for hotspot detection.
I don't know whether sending SCI_SETCHARSDEFAULT each time to clean up will cause a performance penalty.
I'll try to set this up and test shortly, and will send you the results.
Actually, it would be much better to add the additionalWordChar attribute in langs.xml, so that double click word highlighting could take advantage of the functionality as well?
After all the extra oword chrs are part of the language.
This way word chars would be set and reset at lexer change only. See my originally unrelated proposal in idea 769 on IdeaTorrent.
Would you like my implementing that broader picture?
Done. I also sent all the other patches required to compile under VS2013. Its error settings appear to be more stringent than before - perhaps a good thing after all.
Sign up for the SourceForge newsletter:
You seem to have CSS turned off.
Please don't fill out this field.