Re: Modules with files and references to INFDS, but no MAIN, andthe NOMAIN H-spec, was Re: Recursion and CALLB

Mackie, Roger L. (Precision Press) wrote:

I do something like David. As long as the program contains the CLOSE
opcode somewhere, it doesn't matter if it is ever invoked or not. If you
put all the Close <myfile> instructions in a subprocedure you never call
or only call when you want to collect garbage, the compiler won't mind.

Hmm. The complaints about lack of explicit file-close operations were all level 10 warnings.

We also had PGSDS and INFDS issues that were stopping compilation (mainly because the relevant D-specs got moved inside the procedure along with all the other definitions). Can a procedure in a NOMAIN module access the PGSDS? (I'm assuming that INFDS doesn't present a problem).

Also, with regard to opening and closing files, they're opened conditionally on an as-needed, if-not-already-open basis. Will they stay open when the recursions unwind and return to the original caller? Will they still be open when the recursive routine is called again?

The situation at hand is one in which a record in one file might own records in several other files (and those records might own records in other files, and . . .). The object here is to have one call that deletes a specified record in a specified file, then goes through recursively, and finds all the records it might own (directly or indirectly) in other files, until the entire tree that is rooted in the specified record has been deleted.