NetNews Usenet Archive 1993 #1/NN_1993_1.iso/spool/comp/lang/rexx/1465

Text File | 1993-01-07 | 4.2 KB | 107 lines

Newsgroups: comp.lang.rexx
Path: sparky!uunet!csfb1!jbrock
From: [e-mail] (John Brock)
Subject: Re:      Lower-case alphabetic set
Message-ID: <[e-mail]>
Sender: [e-mail] (Usenet News Account)
Reply-To: uunet!csfb1!jbrock
Organization: First Boston Corporation
References: <[e-mail]> <[e-mail]> <[e-mail]> <[e-mail]>
Date: Thu, 7 Jan 1993 16:00:44 GMT
Lines: 94

In article <[e-mail]>, [e-mail] (Jim McMaster) writes:
|> In article <[e-mail]>
|> [e-mail] (John Brock) writes:
|>
|> >In article <[e-mail]>, [e-mail] (John Brock) writes:
|> >|> In article <[e-mail]>,         Wan Hasan Abdullah <[e-mail]> writes:
|> >|> |> Hello netters!
|> >|> |>
|> >|> |> I found that, there's a function to translate all words to upper case
|> >|> |> but I could'nt find in HELP REXX to form the words from upper case to
|> >|> |> lower case by using a function. Could anyone help me to solve this
|> >|> |> problem?
|> >|> |>
|> >|> |> Thanks in advance.
|> >|> |>
|> >|> |> -WAN-
|> >|>
|> >|> How about:
|> >|>
|> >|>   lowercase: procedure
|> >|>   return translate(arg(1), xrange(), translate(xrange()))
|> >|>
|> >|> I like this because it is consise and generic.  It should work as it
|> >|> stands for any language, and it lets me avoid typing out all the upper
|> >|> and lower case letters by hand (which I think looks ugly, even though
|> >|> it's clearly more efficient to do it that way).
|> >
|> >I'm responding to my own post, as there have been other posts
|> >expressing concern about this type of procedure translating
|> >non-alphabetic characters.
|> >
|> >On my system at least, all non-alphabetic characters are unchanged by
|> >the translate function when used with only one argument.  If this isn't
|> >part of the language definition it ought to be!  In any case you could
|> >argue that if c ^== translate(c) for any character c, then c by
|> >definition has a distinct upper case form, and my lowercase function
|> >makes sense.
|> >
|> >It even makes a certain kind of sense if translate is generalized
|> >(e.g., mapping a ==> b, b ==> c, and c ==> a), an long as it never maps
|> >two distinct characters to the same character.
|> >
|> >Here is a procedure which checks which characters on a given system
|> >are affected by the translate function:
|> >
|> >  x = xrange()
|> >  y = ""
|> >
|> >  do i = 1 to length(x)
|> >    c = substr(x, i, 1)
|> >    if c ^== translate(c) then y = y || c
|> >    end
|> >
|> >  say length(y) y
|> >
|> >On my (VM/CMS) system this yields:
|> >
|> >  26 abcdefghijklmnopqrstuvwxyz
|> >
|> >--
|> >John Brock
|> >uunet!csfb1!jbrock
|> >[e-mail]
|>
|> John...this does not work in the general case.  Your "translate(c)"
|> uses the default translate tables, which are designed to translate only
|> lower case alphabetic to upper case alphabetic.
|>
|> The input table "xrange('A','Z')" and output table "xrange('a','z')"
|> are not at all the same.  You certainly will translate "A" (x'C1') to
|> "a" (x'81'), but you also translate the backslash (x'E0') to the EBCDIC
|> superscript character (X'A0').  EBCDIC has a lot of characters
|> (printable and nonprintable) between the letters I and J, as well as
|> between R and S, which will also get translated.
|>
|> Your technique would work in an ASCII system, bacause all upper-case
|> alphabetics are in the range X'41'-X'5A' (with no intermixed
|> characters), and lower-case alphabetics are X'61'-X'7A'.
|>
|> Jim McMaster

I'm sorry, but I don't understand your objection.  What I demonstrated
above is that (at least on my VM/CMS (EBCDIC!) system) the only
characters affected by the translate function are the lower case
alphabetic characters.  So it doesn't matter that my lowercase
procedure translates non-alpha characters, because it translates them
to themselves, leaving them unchanged.  My function does not, as you
stated, change x'E0' to x'A0', at least not on my system.  Try it!  Am
I missing something?

--
John Brock
uunet!csfb1!jbrock
[e-mail]