Re: gawk ignores case with LANG=en_US

[Bob Proulx]

Jim Keniston wrote:
> FWIW, I looked through a gawk manual --
> http://web.mit.edu/gnu/doc/html/gawk_toc.html -- which I found by
> chasing links from gnu.org, and I didn't see anything about this
> behavior.  In particular, I expected to find something about it in the
> "Case-sensitivity in Matching" section.  I note that the manual is dated
> April 1993.  Is there a later version that I should be reading?

Try this one:

  http://www.gnu.org/software/gawk/manual/
  http://www.gnu.org/software/gawk/manual/html_node/index.html
  http://www.gnu.org/software/gawk/manual/html_node/Locales.html#Locales

Or on systems with GNU awk installed you can use the locally installed
documentation.  I prefer the locally installed because then it always
matches the locally installed version of the program.

  $ info gawk
And:
  $ info gawk Locales

> I've been using awk & gawk for maybe 25 years, and this is the first
> time I've noticed this behavior.  I must have done something
> recently to tweak my LANG setting.

In the grand scheme of things locales are new.  Until a few years ago
your systems wouldn't have supported them at all and only the standard
C locale would have been available.  But then as new systems
introduced locale support this behavior would have become available.
On systems this behavior is set in your personal ~/.profile.  But on
other systems it is set for you in /etc/profile or /etc/environment.
Which means that if you were pulling your .profile forward from older
systems that on some you would not have enabled this and on others it
would have.  It has been confusing.  But at this date locales have
become standard to enable UTF-8 and non-ASCII characters which are
important and useful enough to warrant the change in behavior and the
problems that occurs from it.

Bob