Longitude validation (GS1 AI 4309) always fails about postscriptbarcode HOT 7 CLOSED

What implementation of the PS VM are you using?

What does it output when processing the following:

%!PS
/a == (0900000000) cvi ==
/b == 3600000000 ==
/c == (0900000000) cvi 3600000000 gt ==

from postscriptbarcode.

I am on GPL Ghostscript 10.01.2. The output is

/a
900000000
/b
3600000000
/c
false

I encountered the error via treepoem (version 3.23.0) which packages its own barcode.ps. I now built bwipp myself and pointed treepoem to use the barcode.ps under ./build/monolithic_package and ./build/monolithic. With the former, I get the same error, whereas with the latter it works fine. The generated barcode image can also be properly decoded.

from postscriptbarcode.

I now built bwipp myself and pointed treepoem to use the barcode.ps under ./build/monolithic_package and ./build/monolithic. With the former, I get the same error, whereas with the latter it works fine. The generated barcode image can also be properly decoded.

That appears to be a bug in the packaged resource encoder (which I don't maintain). This replaces strings with more efficient binarised tokens. Seems like it's getting the maths wrong for large integers.

If you replace the code in the source .ps file as follows, then rebuild the monolithic_package, does it work?

- cvi 3600000000 gt { pop /bwipp.GS1badLongitude (Invalid value for longitude) false exit } if
+ cvi (3600000000) cvi gt { pop /bwipp.GS1badLongitude (Invalid value for longitude) false exit } if

from postscriptbarcode.

Yes, if I change src/gs1process.ps as suggested and then make again, using the resulting ./build/monolithic_package/barcode.ps gives the same output as with ./build/monolithic/barcode.ps.

from postscriptbarcode.

The packages in release 2023-12-07 now include the workaround.

from postscriptbarcode.

Tested on the latest master 4e130138bddab39193290416841b477e50c9dde8, the generation works fine now 👍

from postscriptbarcode.

Thanks.

from postscriptbarcode.