Re: BREXX DO loop counting weirdness
|
On Fri, Mar 15, 2024 at 6:55?PM Matthew T. Kromer < zerohero0h@...> wrote: It¡¯s a neat bug, in any case.
Yup.? The first time I saw it, I didn't believe my eyes.? And it doesn't always happen, although I've got a good test case now.? This is why my console SPOOL file is always recording now :-)
[...] Curious indeed. I guess we all have our approaches to debugging something like this that's hard to reproduce. I suppose a kind of binary search is the standard start. Suggestions (or questions if you've already done them) that come to my mind are:
What if you assign the value of Cards.0 to a simple variable, and use that as the loop variable instead? (It is interesting, though, that the traced value of the? loop variable I is correct in the trace even when the concatenated result is wrong.)
Similarly, what if you do the concat outside the Say, e.g.
p1 = Fn Ft Fm':' p2 =
'C='I
Say p1 p2? (or even Say p1 and Say p2 separately)
Does anything change with a different input file, in particular, a shorter one - perhaps the first 17 records of the same data?
Your filetype is TEXT, which suggests that it contains non-character data. Of course this shouldn't matter, but it might be good to try a file of the same number of records that is all nice printable chars with no codepage issues (perhaps all alphanum?).
Well I'm sure you can think of this stuff as well as I can. At some point I imagine there'll be a change in the output, and then... But of course this is all the way one debugs Windows software, i.e. with no source code and no real idea of what the program is doing internally.
Keep us posted.
Tony H.
|
