Fabio, I agree that stronger checks during compilation would be beneficial, for example emitting a warning when reserved words are mis-used as function names which renders the functions moot (though this could in fact be intended). However, this is very low on my personal wish list and compilerly things that I would rate higher include:

(1) fix compiler bugs/quirks/misfeatures

(2) true syntax colouring in the editor instead of the current keyword colouring with a bit of lexer level guess-o-matic

(3) ability to turn syntax colouring on for text files and memo fields

(4) make DEBUGOUT work in the runtime environment; it is ideal for producing diagnostic output both to screen (via the output window) or to a log file

(5) disable abbreviations for all newly-added reserved words; the list of reserved words is long enough as it is and enabling abbrs only increases the potential for problems with already-existing user code

... hundreds of others ... ;-)

Note: strong checking during compilation is a two-edged sword, and I'd rather have no checking at all than checks that result in hard errors instead of warnings and that cannot be turned off. And for every warning there should be a means to squelch it right in the source code.

A lint-like facility for Fox would be nice but it could be an external utility written by somebody else, it doesn't have to come from Microsoft. If somebody wants to do something like that as an open source effort then I would be glad to join and I might be able to offer some bits and pieces.

When I started using Fox a couple of years ago I set out to write a linting utility as my 'learn Fox' project. This way I would learn twice as fast, because Fox would be both the means and the subject of the exercise. Since source-code level analysis would have required substantial C/C++ coding I went the object code route instead (I wanted to learn Fox so the code should be in Fox). But extracting semantic information from the compiled code is much more involved than I thought and for obvious reasons you cannot use comments for passing meta info to the linter if the linter sees only object code. So I abandonded the project after I had gotten to the point where it could dump object code as re-compile source code that produces identical byte code output (which is important for automatic verification). Of course, the code is messy and needs to be rewritten from scratch - I was just learning Fox when I wrote the stuff - but it works and so it could be helpful for creating a verification tool for a source code analyzer.