On Mon Aug 1 18:01:43 2022 +0000, Jacek Caban wrote:
While cleanup seems nice, I'm not sure I follow why separating parser and lexer would help testing. I think that testing widl itself is more interesting and it's as simple as adding an interesting IDL file to a regular Wine test. My understanding is that you'd like an easy way to validate test data against midl and you'd like to write midl.exe launcher for that. But that part could be just skipped on Wine, assuming that an equivalent thing happened during the process of building the test itself.
I think that in order for the parser to be more robust, it would be nice to test invalid IDL code too.