I think the differences are largely explainable by different keyboard drivers. I also recently got a cuecat (for scanning books, mostly), and found that it didn't work under X when declawed, only when under the console. (On linux; I didn't try it with windows.) Fornatly, for my application, I wanted to know the type of barcode anyway, and didn't purticularly need human readability, so I simply reclawed it.

The table given in 444574 would tend to be explained by this if and only if we assume that cygwin is doing something mildly odd causing it to use dosish keyboard drivers rather then win32ish ones, which wouldn't much surprise me. (X11 uses it's own keyboard driver, essensially, on linux.)

Since the product was never intended to be used by end-users in "declawed" mode, I suspect the output runs at a frequency which is a little faster then it should be, because it skips the modified base64 encoding step. Some keyboard drivers will still work, some won't, depending on the fine details of the routines.