[vb6] Project Scanner

[LaVolpe]

@Eduardo, wanted to address this one separately. Regarding choosing some validation options and then going back and choosing other ones later, without having to re-scan the project. I agree that would be a good thing, but will have to rework several things. Why?

During initial scan, only the declarations section of each file is fully processed so it can be displayed in the tree. During that processing, critical information is stored with each declaration item so that it doesn't have to be re-parsed during future validations. Procedures are not fully parsed, they are skimmed, and the only information that is stored is the procedure location and size within the code file and whether the procedure header/signature is a multi-line statement, i.e., has underscore continuation characters. Now, when validations are performed, information gathered and written to the recordset are relative to the options selected. The code was written for a one-time pass through the procedures, processing each "word" in the procedure. Most information written to the recordset was intended for one-time use only. For example, parameters are parsed only if needed. The routines expect them not to exist in the recordset at the time they are parsed. If the project were run again, they would be there and foul up the logic. That's just one example of many cases.

What you are asking can be done, and am thinking about it. Just requires additional tracking during procedure parsing to ensure pre-existing flags, attributes, items don't mess up future runs of similar parsing. Also don't want to re-parse something that doesn't need to be