I have made changes to fix the issue and pushed them through to github: * https://github.com/bmTas/cb2xml The jar is available at: * https://sourceforge.net/projects/cb2xml/files/ * Download cb2xml.jar_24_jan_2026.zip
Interesting that Microfocus is different. I tried the leading , with a couple of web compilers (probably GNU Cobol) and got the same result as you did for GNU Cobol
I have made changes to fix the issue and pushed them through to github: * https://github.com/bmTas/cb2xml
The web Cobol compiler I tested the leading comma was not ignored, it was part of the picture
Looking at will make changes as needed
How would you expect 9(6)V99, to be interpreted in a picture clause i.e. 05 CALC-SALES PIC 9(6)V99, VALUE 12.34. Testing using online compilers (GNU Cobol?) the trailing comma is ignored picture while a leading ',' is included. This was raised on cb2xml so I suspect the mainframe also works this way as well so it is likely to standard. If you feel like trying it out here is a sample program IDENTIFICATION DIVISION. PROGRAM-ID. HELLO-WORLD. DATA DIVISION. WORKING-STORAGE SECTION. 01 CALC-COMMISSION-FIELDS....
I have not been able to replicate the issue, I have made changes that I think will fix the issue and uploaded to Git Hub. (https://github.com/bmTas/JRecord) let me know and I will uploaded the latest jars
I will look at the issue and fix. I think I know what the issue is and will issue a fix. But try using IO_STANDARD_UNICODE_TEXT_FILE instead of IO_TEXT_LINE that should solve the issue. In JRecord there are ways to read a Text file As bytes (IO_BIN_TEXT forces this). As Characters then convert to bytes (IO_STANDARD_UNICODE_TEXT_FILE forces this) Constants.IO_TEXT_LINE will choose which ever it thinks is appropriate (basically the encoding). For the most part they work the same, IO_BIN_TEXT solves...