Compiled 4GL vs Interpreted

Neil, RDS is an excellent tool for prototyping and debugging. I have worked at a number of sites where they don't bother comiling at all, they just run the interpreted version. This wa due to the fact that under testing, the RDS versions were only about 10% max slower on average than the compiled versions, and for applications that were interactive, there was no perceivable difference. In general compute intensive applpications will always perform better if compiled, I have no specific figures to support this. Other aspects to consider are: 1) RDS has an excellent debugger, in some versions I have used it could not handle the source being located in many different directories (libraries etc being separated out). This is largely not a problem now. 2) If you are intending to use RDS for debugging and compiling later, they were some differences between code that RDS would treat as valid and code acceptable by the compiler, most notably, the RDS [used to] allows the same BEFORE/AFTER field clause to be used more than once, c4gl produces a compile error. 3) If you organise the source into separate libraries, programs etc, it makes the debugging process more complex. Under RDS because you cat the p-code files together to make the finished program, you can end up with excessively large 'executables' with lots of functions being included unnecessarily. I'm sure that you will get a lot of comments from others on this subject. If you have both available I think it is best to develop and test using RDS. Acceptance test and run compiled programs. Mark Denham BBC London, UK Mark.Denham@bbc.co.uk