On Jun 6, 2008, at 10:51 AM, Dale King wrote:
OK, I see now why it is only cosmetic. It only ends up being that way
because the strings differ only in a character that is not a letter.
That is a cumbersome way to do case insensitive comparison. It would
be a lot easier to lose the third parameter and do:
if (arg1.size() != arg2.size() ) return false;
for(int i = 0; i < arg1.size(); i++) {
if (apr_toupper( arg1[i] ) != apr_toupper( arg2[i] ) ) return false;
}
(using apr_toupper for maximum portability). This could have problems
if you were trying to compare against accented or foreign characters,
but I think you are only comparing against ASCII.
That approach could have problems with the letter "i" due to the
Turkish i problem (http://en.wikipedia.org/wiki/Turkish_dotted_and_dotless_I
) which did affect log4j (https://issues.apache.org/bugzilla/show_bug.cgi?id=40937
). It seemed easier to just avoid the whole locale sensitivity of
case changing strings that contain "i" using the RTL and just hard
code the expected uppercase and lowercase versions.
