Re: 4GL Conversion

The biggest problem with Compiled Informix to Interpreted Informix was the fact that you MUST use any return values from a function. If your function returns a value, calling it with "CALL" instead of an assignment value will result in an error.... not a good thing, given the code we had to port over. (I know about CALL... RETURNING, but you get my point.) Also, the whole business with the runner is a pain to maintain and is buggier than any other Informix product we use (believe it or not, you can't compile ANY runner under version 4.0 on our platform. I had to ask the tech support people if they even tried to RUN the program before they distributed it). We can't set "int_flag" or get any routine that uses "set_break()" to compile into the runner (though we have a workaround for both problems). <Shrug> The runner and I do not get along too well..... The program compiler is busted; it doesn't always compile out-of-date modules if you keep different modules in different directories. We have made a pretty complex Makefile to handle both the compiled, interpreted, and runner versions of the Informix programs, and a simple program to build fgiusr.c for you automagically, which made up for a lot of its shortcomings. All other things aside, I would HATE to go back to developing Informix applications only under the compiled version. Even with its quirks, RDS is an amazingly good production booster. Lee Brintle | ``And so, I leave you with this final word: Advantage Information Mngmnt | twang.''