Re: Case insensitivity seems to ignore lower bound of interval

[Eric Bischoff]

Le mercredi 27 avril 2011 20:48:41, Aharon Robbins a écrit :
> Greetings. Re the below.
> 
> First, thank you for the bug report.
> 
> Second, it's not a bug, but rather the consequence of how locales behave.
> This is documented somewhat in the released gawk manual and documented
> better in the upcoming one.

Hi Aharon,

First, thank you for your kind answer.

I am really sorry, but I think you did not read my bug report accurately. It 
looks like you gave the answer to another question than the one I asked.

What you are answering here is a Frequently Asked Question about un undesired 
behaviour of the locales, but this is not what I am speaking about.

What I am speaking about is not documented. On the contrary, I showed you 
where the behaviour contradicts the documentation.

Let me restate my bug report, perhaps it was not clear enough.


1) Contradiction with the documentation :

http://www.gnu.org/software/gawk/manual/gawk.html#Locales says that

     $ echo something1234abc | gawk '{ sub("[A-Z]*$", ""); print }'

returns

      something1234

while it returns

     something1234a

Please notice the *a* at the end. That's what this bug report is about. This 
"a" has no reason to appear.


2) Another way to reproduce :

$ echo "ijklmnopqrstuvwxyz" | awk '{ gsub(/[R-Z}/, "X"); print }

ijklmnopqrXXXXXXXX

Please notice the "r".

Why does the "r" not be changed into "X", while "s", "t", "u", "v", "w", "x", 
"y", "z" do?


I think this is a real bug.

The fact that the locales are taken into account in the intervals is indeed 
questionable, but is another topic.

Sorry to be insisting. I still hope that helps.

-- 
Éric Bischoff - Bureau Cornavin
Technical writing and translations
http://www.bureau-cornavin.com
(+33) 3 68 46 00 85
sip:address@hidden