INITIAL COMMENT (WAS: 
Author Message
 INITIAL COMMENT (WAS:

Hi folks,

sorry for the double, and triple, quotes. They are inevitable, as the
poster I'm replying to has not kept the original senders' identities.

On Tue, 24 Jan 1995 19:30:49 +1100 Ian Timms said:

Quote:
> > > S1 ?== S2 invalid
> > What exactly does "valid" and "invalid" mean?
> It means that the equivalence test _really_ did test true.

The simple observation that
      say S1 ?== S2
will output
      0
does not help much to understand the issue.

Quote:
> > > [...] Then on further finding that '?' and '!' are valid symbol
> > > characters and also unninitialised [I] would have expected REXX to
> > > replace them with the uppercase versions of themselves [...]
> > > Instead it just treats them as though they weren't even there at
> > > all,

REXX does not treat them as though they weren't there. Rather, it
evaluates the expression containing them (as everybody should have
expected). In the example at hand, this expression will yield 0.

Quote:
> > > when it should really have the sense to pull an error.
> >  It is correct syntax, so it should not give an error.

You can ask REXX to give an error on any access to an undefined variable:
     signal on novalue
Without this proviso, REXX indeed is not allowed to issue an error, when
the syntax is correct.

Quote:
> > Have you tried it with "trace i" on?
> Nope, I was testing possible equate options, not debugging variables,
> so no I did not bother using trace.

In REXX, use
  Trace I
to explicate, stepwise, how an expression is evaluated. Do never let your
prejudices prevent you from using the most effective tools available!

Quote:
> I fully expected such things to raise an error condition,

Such things? What sort of things? A syntacticaly valid clause? Raise an
error? Get real, Ian!

REXX has been designed by Mike C, not by Ian T. This design comprises,
amongs others, the following features:
- question marks, exclamation points, and underscores, can be used in
  symbols just as letters can,
- no intermediate blank is needed to delimit tokens (as long as no
  ambiguity arises),
- in expressions, a space can denote an operator, viz. concatenation
  with a space inserted.

Hence, the expression
      S1 ?== S2
is parsed as:
      ( variable(S1)
        operator(concatenate-with-space)
        variable(?)
      )
      operator(equals)
      variable(S2)
This expression is evaluted as explained in a previous poster, and as
easily observed with TRACE I.

Quote:
> I still think the use of '!' and '?' in the context described above
> _should_ raise an error, _particularly_ if they have been reserved for
> future use

They haven't.

Quote:
> >  I sincerely doubt that "!=" means "not equal" in any interpreter.
> There are a number of C interpreters, not to mention shell scripts

REXX is not C.

With best wishes from your REXXpert,

*** Please use only my new address at uni-konstanz.de, as all Bitnet
*** addresses at DKNKURZ1 have expired, and all Internet adresses at
*** Nyx.Uni-Konstanz.de will do so some time in 1995.



Sun, 13 Jul 1997 02:22:57 GMT  
 INITIAL COMMENT (WAS:

[about a character that appeared in his article]

Quote:
>Logical 'not' character which is a graphic in the standard PC character
>set, the little top right corner box type one. (obviously gets translated
>in transmission/display somehweres).

Obviously the PC character set is not the one that everyone in the Unix
world uses. :-)  This: ? is a logical "not" character...

[about x != y]

Quote:
>                                    I fully expected such things to raise
>an error condition, I still think the use of '!' and '?' in the context
>described above _should_ raise an error, _particularly_ if they have been
>reserved for future use and are therefore not really suitable for usage
>as symbols in current rexx implementations.

As far as I am aware, ? and ! always have been and always will be valid
symbols in Rexx, and are not reserved.  I see no reason for "x != y" to
raise an error, any more than there is a reason for "x z= y" to raise an
error.

i>          you might run into problems if you decided to name your variables
i> in a Z-like convention (input variable names end with "?" and output
i> variable names end with "!").  What does

i>         result!=3

i> mean in that situation?...

Quote:
>What is 'Z-like convention', I have never heard of this although
>I have a funny feeling it harks back to some BASIC thingy.

On the contrary!  Z is a specification language in formal computing
science (which was developed at Oxford University and has been used in
the development of IBM's CICS).  I notice you didn't answer the question
though.  :-)

Ian Collier - Departmental Lecturer (and perpetual postgrad student) -

Building, Parks Road, Oxford OX1 3QD - WWW Home Page (including REXX
section): http://www.comlab.ox.ac.uk/oucl/users/ian.collier/index.html

New to this group?  Answers to frequently-asked questions can be had from
ftp://rexx.uwaterloo.ca/pub/rexxfaq.txt .



Mon, 14 Jul 1997 19:03:25 GMT  
 INITIAL COMMENT (WAS:

Quote:



> [about a character that appeared in his article]
> >Logical 'not' character which is a graphic in the standard PC character
> >set, the little top right corner box type one. (obviously gets translated
> >in transmission/display somehweres).

> Obviously the PC character set is not the one that everyone in the Unix
> world uses. :-)  This: ? is a logical "not" character...

                         ^
This has come out here as 1/4 (x'AC')! Furthermore it was different in the
REXXLIST version, I can't remember what but it wasn't recognisably a logical
'not'. I'm using OS/2 Warp/Kedit with news coming through UUCP.

Quote:
> [about x != y]
> >                                    I fully expected such things to raise
> >an error condition, I still think the use of '!' and '?' in the context
> >described above _should_ raise an error, _particularly_ if they have been
> >reserved for future use and are therefore not really suitable for usage
> >as symbols in current rexx implementations.

> As far as I am aware, ? and ! always have been and always will be valid
> symbols in Rexx, and are not reserved.  I see no reason for "x != y" to
> raise an error, any more than there is a reason for "x z= y" to raise an
> error.

FWIW ? and ! being valid symbol chars has always seemed odd to me and on
reflection it's because in natural language they're never part of words
let alone names, except in a few unusual cases like 'Westward Ho!'.

Yours, Patrick
_____________________________________________________________________________


I tend to eat my UUCP feed once a day so replies can take two days



Tue, 22 Jul 1997 12:12:14 GMT  
 INITIAL COMMENT (WAS:

Quote:


>> Obviously the PC character set is not the one that everyone in the Unix
>> world uses. :-)  This: ? is a logical "not" character...
>This has come out here as 1/4 (x'AC')!
>       I'm using OS/2 Warp/Kedit with news coming through UUCP.

Which proves my point, because you are using a PC and you are not using the
same character set that everyone in the Unix world uses.  :-)

Try an ISO8859 (ISO LATIN 1) character set...

Ian Collier - Departmental Lecturer (and perpetual postgrad student) -

Building, Parks Road, Oxford OX1 3QD - WWW Home Page (including REXX
section): http://www.comlab.ox.ac.uk/oucl/users/ian.collier/index.html



Fri, 25 Jul 1997 20:18:05 GMT  
 INITIAL COMMENT (WAS:
On Mon, 6 Feb 1995 12:18:05 GMT Ian Collier said:
Quote:



>>> Obviously the PC character set is not the one that everyone in the Unix
>>> world uses. :-)  This: , is a logical "not" character...

>>This has come out here as 1/4 (x'AC')!
>>       I'm using OS/2 Warp/Kedit with news coming through UUCP.

>Which proves my point, because you are using a PC and you are not using the
>same character set that everyone in the Unix world uses.  :-)

>Try an ISO8859 (ISO LATIN 1) character set...

Code page 1004 has the logical not in the correct position for
windoze or OS/2. 1004 is "almost" ISO8859 with extra characters.

--------

Systems Programmer -- Team OS/2
NYS Legislative Bill Drafting Commission
1450 Western Avenue, 3rd floor
Albany, NY 12203
Voice:(518)458-5114  FAX:(518)458-5108

"Help stamp out slogans in signature files"



Fri, 25 Jul 1997 18:59:18 GMT  
 INITIAL COMMENT (WAS:

Quote:

> On Mon, 6 Feb 1995 12:18:05 GMT Ian Collier said:


> >>> Obviously the PC character set is not the one that everyone in the Unix
> >>> world uses. :-)  This: , is a logical "not" character...

> >>This has come out here as 1/4 (x'AC')!
> >>       I'm using OS/2 Warp/Kedit with news coming through UUCP.

> >Which proves my point, because you are using a PC and you are not using the
> >same character set that everyone in the Unix world uses.  :-)

> >Try an ISO8859 (ISO LATIN 1) character set...

> Code page 1004 has the logical not in the correct position for
> windoze or OS/2. 1004 is "almost" ISO8859 with extra characters.

OS/2 doesn't support code page 1004; all its pages except the Arabic one have
the fallen L at x'AA'. Your reply has translated the logical 'not' to a comma
(x'2C'); I'm not going to switch code pages depending on whether the text has
come through a particular Net hub (or whatever it is that's translating). :-)
We really do need the greatest lower bound.

Yours, Patrick
_____________________________________________________________________________


I tend to eat my UUCP feed once a day so replies can take two days



Sat, 26 Jul 1997 23:08:40 GMT  
 INITIAL COMMENT (WAS:

Quote:

>As I always predefine everything I am going to use, I normally have no
>cause to use 'signal on novalue'  [ ... ]

Ahem - that is EXACTLY the circumstance under which you SHOULD use
"signal on novalue"!  If you habitually predefine everything, you
want to catch any and all errors from inadvertently misspelling
symbols or using characters like ! and ? in your source.

If, on the other hanmf, you habitually relied on undefined
symbols representing themselves, you would not want to use
"signal on novalue" precisely so as to enable that sort of
behavior, and you would be opening yourself up to the{*filter*}
case of ? and ! being interpreted as variable names that
equate to themselves.

 - seb



Wed, 30 Jul 1997 03:57:08 GMT  
 
 [ 7 post ] 

 Relevant Pages 

1. Checking Undefined Variables (was: INITIAL COMMENT)

2. Initial Comment (Was:

3. Initial Comment

4. INITIAL COMMENT (WAS:

5. Java and Eiffel - Further comments: I am depressed.

6. I am not deaf, but am I mute?

7. Counting commented lines (full-line comments only)

8. comments on comments

9. meta-comments for comments on proposals

10. comments on comments

11. comments on comments

12. Single Pound # comments versus Double Pound ## comments

 

 
Powered by phpBB® Forum Software