Adrian Sutherland
Adrian Sutherland
I've pushed a potential compiler fix for this issue - @rvjansen @Peter-Jacob can you retest it?
> I just saw the changes you made to left() to swap back registers: > ``` > swap r4,a1 /* Swap back - pedantic */ > swap r6,a2 /* Swap...
> **In any case, for human readability I think good practice should be to swap them back although there is a performance penalty** > This is actually the reason I...
Well, the answer is both simple and complex. The easy answer is that the 3 swaps at the end are not necessary - and are not necessary in general. However,...
> * The example you used passes a simple string, so it's not an array problem, is it? The specific issue was with linking and swapping registers and not "unwinding"...
You can remove the swaps I added to left() Also I think if we make left a standalone procedure, not calling substr(), and written in REXX it might be faster...
Can you post the dotest.rexx file - thanks
In f0043 we have improvements in that the compiler doesn't just die! This file (I commented out 8 lines from line 60 - stems): ``` /* Test REXX Program -...
Should we leave this for level c?
Let me revisit when I have the options decimal - I will try and make the code "levelb" friendly ...