What is the latest Aztec specification?

[Joe O'Rourke]

I'm having mixed results with my C++ implementation of encoding strings into Aztec symbols.  I am using
AIM International ITS/97-002 dated 1997-11-05.  The spec clearly states that if a codeword needs to be padded
to fill it out to the required size, then it should be padded with ones 1's UNLESS doing so results in ALL ONES
in which case it should terminate in a final zero.  When I do this, however, the symbol will not scan. I've
reversed some web-generated symbols that DO scan and they end the string with '111101'  instead of '111110'.
This is in clear violation of my spec but they work!  I suspect I might be working with an outdated spec.  Can
anyone help me?

[Sean Owen]

I think that's very old... AFAIK the latest is the ISO standard:

ISO/IEC 24778:2008

http://www.iso.org/iso/catalogue_detail.htm?csnumber=41548

[Joe O'Rourke]

Thanks Sean,
I got the new ISO spec.  It did, in fact, add a caveat about adding pad characters.  However, it was
very vague on how to fix it.  It essentially said to avoid the occurrance, while padding, of a Byte Shift
character '11111' = 31.  I reverse engineered some other's code and they terminate the last codeword
as '11101' to avoid the B/S but the spec doesn't say to do this.  My problem is this: Where in the
string does a scanner look for the '11111' pattern?  For example, if I have a 10-bit codeword that ends
in '101001XXXX' where X=pad character, should I fill it as '1010011111' or fill it as '1010011110"?
Best,
Joe

[Sean Owen]

I don't know myself, I don't have the spec and didn't write this part of the code. I would follow the behavior you see in the wild if there is an ambiguity. I'm sure the code proceeds here by parsing symbol by symbol. If there are too few bits left to read a new symbol I assume it just stops.