epicsStr[n]CaseCmp()

[Andrew Johnson <anj@aps.anl.gov>]

epicsStringTest.c currently proves (amongst other things) that

    testOk1(epicsStrnCaseCmp(" ", "", 1) < 0);
    testOk1(epicsStrnCaseCmp("", " ", 1) > 0);
    testOk1(epicsStrnCaseCmp("abcd", "ABCDE", 1000) > 0);
    testOk1(epicsStrnCaseCmp("abcde", "ABCD", 1000) < 0);

    testOk1(epicsStrCaseCmp("abcd", "ABCDE") != 0);
    testOk1(epicsStrCaseCmp("abcde", "ABCD") != 0);
    testOk1(epicsStrCaseCmp("abcde", "ABCDF") != 0);

Unfortunately the first group are wrong, and glibc's str[n]casecmp()
functions on RHEL6 agree with me. Longer strings should always compare
greater than any initial subset of themselves. The second group probably
uses != because the author didn't want to think about and fix the bug at
the time...

Any comments?

- Andrew
-- 
Advertising may be described as the science of arresting the human
intelligence long enough to get money from it. -- Stephen Leacock