Re: When is "z" != "z" ?

At 12:27 PM -0700 6/4/06, Rasmus Lerdorf wrote: >tedd wrote: >>But, what brothers me about the routine, is that is DOES print "z" where it is >>supposed to. In other words, the characters a-z are output before continuing with aa and so on. The >>operation doesn't end with "z". > >Your condition for the loop to continue is $i<="z". > >[1] When $i = "y" it will obviously continue because "y" < "z" >[2] When $i = "z" it will obviously continue because "z" = "z" >[3] When $i = "aa" it will obviously continue because "aa" < >"z" I agree with [1] and [2], but [3] is where we part company. You see, if you are right, then "aaa" would also be less than "z", but that doesn't appear so. >It doesn't stop until you get to "z"+something. As in "za" because at >that point "za" > "z" so the last thing you see is the one before >"za" which would be "yz". See above -- plus, I would expect that in evaluating a character, that character(s) would not enter the mix. If I was to evaluate a single digit, it would seem odd if double digits were allowed in that evaluation. >>Here's another way to look at it. All characters before "z" are less than >>"z" -- correct? So, what value are all characters after "z" (i.e., >>"aa-yz")? They cannot be greater than, nor can they be less than. So, what are they? > >But you are not comparing things in the same context here. Strings are naturally compared >alphabetically while you are thinking they are compared numerically somehow. Think of sorting a set >of words. Would you expect "aa" to sort before or after "z" ? Not meaning to infer that you don't know this, because I know you do -- so I must be confused. Yes, it is my contention that strings are numerical -- you don't store "A" in memory, you store 0100 001, or ASCII DEC 65. Likewise "a" is DEC 97 (0110 0001) and "z" is DEC 122 (0111 1010) and if I compare "a" to "z" , it will always be less by numeric definition. However, if I compare "aa" with "z", then what? Is the numerical translation of the strings -- 0110 0001 0110 0001 < 0111 1010 equal to TRUE -- I don't think so. Likewise, is the string "aaa"" greater than "z" -- yep, I can buy that. But, if you hold that "aa" is less than "z", then everything must be less than "z" (except z stuff). So, the loop would continue until you hit the maximum allowable for strings -- that doesn't sound good. I can't see the logic that says "aa" is less than "z" while "aaa" is greater. That sounds like something our government would dream up. >So yes, like I said, it is a bit of a quirk, but there is no good answer to what "z"++ >should be and we certainly don't want to change the default comparison mechanism to not compare >strings alphabetically because that would screw up all sorts of stuff including usorts and >peoples' expectations. It's just in this case where you have gotten it into your head >that incrementing strings is a good thing to do. Oh, I'm not saying to change anything. I've seen and worked with worse. >You'd be much better off with a range call to quickly generate a range of characters. You >could then loop through that character by character. Or use the chr() function to work with >character codes instead. I figured a way around it, I just thought it was odd. Thanks for your insight. tedd -- ------------------------------------------------------------------------------------ http://sperling.com http://ancientstones.com http://earthstones.com