Add support for JAB Code

[terryburton]

TL;DR Production of these symbols to too computationally expensive (and therefore slow) for a PostScript implementation. Basic symbols would take in excess of a minute to generate.

Specification: https://www.bsi.bund.de/EN/Publications/TechnicalGuidelines/TR03137/BSITR03137.html

[terryburton]

Preliminary code is available with the following limitations:

• No optimised high-level data encoding.
• Master symbols only.
• Too slow to be usable due to the high computational complexity of the LDPC error correction as defined at present.

https://github.com/bwipp/postscriptbarcode/blob/master/src/jabcode.ps

Development is stalled waiting for possible changes by the JAB Code authors.

[terryburton]

On hold as the specification is in a state of flux. Working with the upstream authors on robustness and standardisation related issues.

[hvitis]

Any progress on that? How slow would it be to implement error correction with JS for JAB code atm? Is it feasible?

[terryburton]

The draft specification that is being worked on by ISO/IEC JTC 1/SC 31 WG 1 is structurally distinct from the published German BSI specification.

The LDPC process requires ~10,000,000 iterations of the innermost loop which is unlikely to be efficient in any interpreted runtime. A simple transliteration of the C implementation to PostScript yielded the following timings for a relatively small 8-colour JAB Code symbol:

• 1420ms createMatrixA
• 11200ms GaussJordan
• 250ms createGeneratorMatrix

I would be surprised if a JavaScript JIT were to accelerate this a great deal.

[hvitis]

I understand. Thank you very much!

[terryburton]

Closing. If someone whats to complete the implementation against the ISO standard then they are welcome to, however it seems that it would be practically unusable in a PostScript environment.