[SOLVED] DSRN (dog slow roman numerals)

[HP67]

(06-12-2014 12:23 PM)Dave Britten Wrote:  

(06-12-2014 11:44 AM)HP67 Wrote:  Are you talking about inside the CASE statement where the debugger appears to be stepping over all the unselected clauses one by one? If so, I noticed this and I don't understand it. Obviously CASE does work, and once a clause is selected the other clauses are certainly not selected. But sometimes (always?) the debugger appears to be testing all the clauses when you STEP.

Yeah, the CASE statement would have to evaluate each condition sequentially until it finds the correct block to execute. As the remaining value gets smaller, you'll see it taking longer to work its way down to the bottom conditions. The 48 CASE statement is really more if an IF/ELSE IF/ELSE structure than the case structure from C, etc. In other words, you're getting a lot of extra DUP >= checks when the remaining value is small.

Oh, that's not what I meant. When debugging CASE statements in other code, the debugger seemed to be evaluating other clauses after the clause that was selected. It's easy to prove the code doesn't do that but in the debugger it sometimes looks like it does.

As far as what I think you're talking about now, having to DUP DUP DUP did bother me because it doesn't look very sano. But at the same time DUP is cheap, probably one of the cheapest ops in RPL so in practice it shouldn't do anything bad. And, if the code has to check for a certain condition, CASE and IF/THEN/ELSE are all the same. There are always going to be a bunch of redunant checks until you get to condition you want. The only way around it would be to code a very complicated CASE or IF/THEN/ELSE that would repeat only relevant tests. I thought about it, but didn't want to try it badly enough to actually code it for this.

Thanks Dave, Thomas for your comments and code.