Mercator 6.7 bug "C/A"

[BJZeak]

Very strange bug popped up after 10 years of active use ... has anyone seen other issues like this?

Mercator Map INPUT card received a STRING that started with "C/A" and MAPS to an OUTPUT Card ANSI TEXT Field that has only one restriction that being MAX number of characters set to 80 ... both type tree's have no other rules or restrictions surrounding either of the fields involved.

Mercator mapped this string with out any errors or warnings as:

INPUT: "C/Aabcdefghijklmnopqrstuvwxyz"
OUTPUT: "abcdefghijkl"

Basically Mercator threw away the C/A and then truncated the remainder of the string to 12 Characters (C is 12 in HEXIDECIMAL not sure if that has anything to do with it)

I played around with this a bit further to qualify what the trigger was and it only appears that "C/A" is causing the issue:

Is the issue caused by "/"? This result by itself might suggest that ...
INPUT: "C-Aabcdefghijklmnopqrstuvwxyz"
OUTPUT: "C-Aabcdefghijklmnopqrstuvwxyz"

If C is some value to truncate the line with a /A then would D increase the number of characters ...
INPUT: "D/Aabcdefghijklmnopqrstuvwxyz"
OUTPUT: "D/Aabcdefghijklmnopqrstuvwxyz"

This issue appears to be case sensitive as well
INPUT: "c/Aabcdefghijklmnopqrstuvwxyz"
OUTPUT: "c/Aabcdefghijklmnopqrstuvwxyz"

INPUT: "C/aabcdefghijklmnopqrstuvwxyz"
OUTPUT: "C/aabcdefghijklmnopqrstuvwxyz"

 

[eyetry]

would need to see your type trees

 

[BJZeak]

Sorry!!! There is NO issue with Mercator this was an ID 10 T issue. sigh

There were several instances of the same segments with 2 identical segments in the same loop ... I traced backwards from the first segment and had formed my conclusions based on the test results and rule searches both in the type tree and map ... the 2 segments in question could either export together, one or the other, or not at all. It was the one or other that caught me. The first instance had no rules which I read as being the issue ... the second instance had the rules which were causing the strange behavior. We had more failures over night that finally gave away the issue because the 2 instances finally showed up in the output file together.

I have now fixed the original problem