Re: m4 ebcdic patch

[Michael Elizabeth Chastain]

Hi Paul,
> It's in the regex code that looks for patterns like '[a-z]'; it
> compares for characters greater than or equal to 'a' and less than or
> equal to 'z' in the collating sequence.

I took care of this.

My patch fixes the code in regex.c:compile_range which expands ranges
to character sets.  I call my new function make_range, which works
with any character encoding.

I also wrote new tests, like these:

  regexp(`~{}\mec.was.here', `[A-Za-z'])
  dnl @result{}4

  patsubst(`Alpha_{}\~-bet.', `\w', `& ')
  dnl @result{}A l p h a _ {}\~-b e t .

In the ebcdic table that I have, '~' falls between 'r' and 's', and '\'
falls between 'R' and 'S'.  These tests show that they are not
matching [A-Za-z] and not matching \w.

Of course I could have missed something.  Can you point to something
that I missed, or suggest a test case that you want to try?  I don't
have access to an ebcdic machine, but I am happy to add more test cases
and ask Kevin Mullins to run the test suit on his s/390.

Michael C