MatchPatternNoCase(). But it probably effects all Match#? dos functions.
Ah, yes. I remember a discussion about this on the Dev list. The problem is that the autodoc never was really very specific about the error code ("Result: Success or Failure") and was supposed to return DOSTRUE in the case of success, which is -1. OS 2 and OS 3 returned 1 instead, which is, technically speaking, wrong.
That's another example for the unfortunate BCPL legacy.