On 2018-02-17, at 08:24:38, Robin Vowels wrote:
>
>>>> The prior declaration is (almost) necessary for single-pass compilation.
>>> No it isn't.
>>> XPL is a single-pass compiler, and declaration of identifiers
>>> is not required.
>> I said "almost".
>
> "Almost" was in parentheses.
>
> Whether almost necessary or necessary, it is not true.
> Declarations of labels are not required.
>
I've worked maintaining and extehding a Pascal compiler. If the
requirement for label declarations were removed some small amount
of additional code and coding would have been required when both
were expensive. But it was probably a false economy to shift a
cost from the supplier side to the consumer side.
And earlier Pascal compilers had a loophole where an identifier
in an outer scope might be referenced prior to its declaration
in an inner scope. I added a check, calling that a syntax error.
Some languages consider the scope of an identifier to go from
the point of its declaration to the end of the block -- prior
to that declaration the identifier in the outer block may be
used.
On 2018-02-17, at 08:31:57, Robin Vowels wrote:
>
>> With suitably unrestricted identifier length, alphanumeric labels
>> have more evocative value than numbers.
>
> With names, you don't know where to look (up or down).
>
> With integers, provided that they are arranged in increasing order,
> you know where to look.
>
o One might adopt a convention such as "L0123_process_transaction:"
to keep labels in lexical order and retain evocative value.
o Most development systems, even mere editors, make it easy to
find any occurrence of a symbol.
o Do Assembler programmers find much value in keeping labels in
lexical order? I know that seems to be the custom more in
conditional assembly than in executable code.
o Seriously?
I once had a supervisor who, based on experience in previous employment,
insisted that all our FORTRAN subroutines have names incorporating a
revision level, such as:
SUBROUTINE A123V45
C (Use this name in CALLs.)
ENTRY PROCESS
...
RETURN
END
The subroutine name appeared in the load map; the entry name did not.
Help desk personnel were bemused when I went to them with a problem
and showed them a map ("Hey, Nick, look what this guy's manager made
him do!")
> Rarely do I use FORMAT statements in any language.
>
FORTRAN (formerly?) required them.
-- gil
