scott moore

I'll take the cases one by one. The quoted article
is from my site:

http://www.moorecad.com/standardpascal/ansiiso.faq


(1) This is absolutely incompatible with J&W (AKA 1972 revised Pascal).

(2) The standard extended the original J&W in this case. Fully compatible.

(3) Incompatible.


(4) Compatible. J&W 1972 left these rules up to the implementation. An
implementation of ISO 7185 would be compatible with J&W 1972 for this case.


(5) Compatible. The key, again, is "implementation defined". An ISO 7185
implementation would be able to run J&W programs that didn't rely on
short labels, as J&W says not to.


(6) Compatible. Again, implementation defined.

(7) Incompatible, because a valid J&W program, using a for loop variable
defined in a different block, would not compile, and it wasn't labeled as
"implementation defined".

To summarize the rules that would PREVENT a valid J&W 1972 program from
running on a ISO 7185 implementation:

1. Use of procedure/function parameters without also declaring their
parameters.

2. Passing standard system procedures and functions as procedure/function
parameters.

3. Use of for loop variables from a different block.

These are hardly common cases, and they are quite easy to change. Further,
Wirth himself requested the procedure/function parameter change. Finally,
J&W, in the third and forth edition, fully incorporated these changes.

Certainly it has become common to spread the myth that J&W 1972 is
incompatible with ISO 7185, but the facts say different. The key is being
more interested in the facts than the retoric. If you are trying to "prove"
that ISO 7185 is incompatible with J&W, you can find examples of programs
that won't compile under ISO 7185 Pascal. But if you look at J&W's rules
of 1972, you will find that the same programs were disallowed under
that standard as well, since they use things like gotos to nested structures
and other constructs that J&W 1972 fully allows implementations to reject
(which is the meaning of implementation defined).

Scott A. Moore

2metre

I presume this is a continuation of the earlier thread (Re: Current test
version of Chess 0.5) that had diverted into a discussion over whether
ISO 7185 Pascal is the Original Pascal.

This discussion was prompted by Scott Moore writing:
"The standard IS Pascal, its what Wirth introduced as Pascal."

I (Mike Price, aka 2metre) corrected Scott Moore, pointing out that ISO
7185 was *not* the orginal.

In my penultimate post on this topic in that thread I asked the question:

"Is the ISO standard identical to Wirth's original specification? (eg as
described in his paper "The Programming Language Pascal" - or the
revised edition).
If the answer is "YES", then I will accept that the standard is, for all
intents and purposes, the ORIGINAL.
If the answer is "NO", then the ISO standard IS NOT the original."

Scott Moore then suggested that this topic should be reconvened in a
new, more suitable, thread.

Scott Moore has started this thread, but asking a different question.
However, the wording of the thread title seems to confirm my original
assertion: it differentiates 'ISO' Pascal from the 'original' Pascal.

I find it interesting that Scott now asks the question "Is ISO Pascal
*compatible* with...". He then invokes J&W which (AFAIK) was published
in 1974. In my opinion, Pascal can be traced back to the earlier
publications by Wirth "The Programming Language Pascal" (June 1971) and
"The Programming Language Pascal - The Revised Report" (Nov 1972). I
have not researched the differences in these two publications and J&W,
so I don't currently take issue with Scott's current assertion that J&W
is the original.

The reason I first questioned whether ISO Pascal is the original Pascal
is because I felt there was an implication in Scott's statement that any
language that does not conform to the standard does not earn the right
to be called Pascal.

Mike

2metre

Now I'll move (after a coffee break and a few bits of proper work...) to
the question you've posed.

Some clarification:
Would you consider using the phrase 'conform to' in place of
'compatible with'? It would then be easier to directly relate your
statements to the standards.

For further clarification:
When asking the question " Is ISO Pascal compatible with J&W (original)
Pascal ?" do you mean
"Is an ISO Pascal Processor compatible with a J&W program"
or vice-versa?
Unless you wish to make a distinction, then for the two to be compatible
with each other, both versions of the question must be satisfied.


Whether they are easy to change or not, they are incompatibilities.
Whether they are common or not, they are incompatibilities.

Now I am somewhat confused. You very elegantly analysed the differences
and came to the conclusion that incompatibilities exist. You then
appear to totally ignore your own conclusion and state:

You have very nicely shown that the facts AGREE with the 'myth'.

To quote an earlier post of yours:
"No, I don't expect any of these facts to impinge on your discourse,
which dosen't rely on facts in any case."
I am relying on the FACTS that you have presented.

scott moore

Just one slight correction. I never said the above.

--
Samiam is Scott A. Moore

Personal web site: http:/www.moorecad.com/scott
My electronics engineering consulting site: http://www.moorecad.com
ISO 7185 Standard Pascal web site: http://www.moorecad.com/standardpascal
Classic Basic Games web site: http://www.moorecad.com/classicbasic
The IP Pascal web site, a high performance, highly portable ISO 7185 Pascal
compiler system: http://www.moorecad.com/ippas

Good does not always win. But good is more patient.

scott moore

The standard clarifies the rules of J&W 1972. A implementation
that used ISO 7185 would qualify as an implementation of J&W 1972.

Note that you must use "J&W 1972", because:

1. In 1969 the language Pascal was created.

2. In J&W 1972 (the "report", language was revised. See my web site, it
has that original copy online.

3. In 1983, J&W was revised again to meet the ISO 7185 standard.

The language you would call Wirth's standard Pascal, is actually
the langauge as of 1972, as reported by J&W.

Much as with the ANSI C standard, the purpose of ISO 7185 was not to
change the Pascal language, but to clarify it.

Because a program (as has been seen) could easily have been written to
be dependent on the CDC 6000 version of Pascal, it should be generally
agreed that the test of if a program runs or does not run from the
original CDC 6000 implementation to other implementations is not a
test of the language Pascal, otherwise all Pascal implementations would
be considered different. So the question "does an implementation
qualify as Pascal" must be performed by comparing the rules of J&W 1972
to the rules of ISO 7185. The question "is ISO 7185 Pascal different
from J&W 1972 Pascal" can only be answered by critically examining if,
and how, it would be possible to write a program in J&W 1972 Pascal
that was valid, conformed to all stated machine dependent rules (as
again, stated by J&W 1972 Pascal), but would not also be processed
correctly by ISO 7185 Pascal.

First, its clear that you made up your mind about this subject long
before reading this thread. That changes the "discussion" to an argument,
or at least a rebuttal.

I have answered the question "is ISO 7185 Pascal" the same as Wirth's
Pascal, so my part is done. Feel free to use this space as you desire.

--
Samiam is Scott A. Moore

Personal web site: http:/www.moorecad.com/scott
My electronics engineering consulting site: http://www.moorecad.com
ISO 7185 Standard Pascal web site: http://www.moorecad.com/standardpascal
Classic Basic Games web site: http://www.moorecad.com/classicbasic
The IP Pascal web site, a high performance, highly portable ISO 7185 Pascal
compiler system: http://www.moorecad.com/ippas

Good does not always win. But good is more patient.

2metre

Agreed. I did say that *I felt* there was an *implication*. My statement
was very strongly qualified both as a feeling and an implication.

By stressing that "The standard IS Pascal" (your caps, not mine) you
(IMHO) implied that anything else is not Pascal.

2metre

1/ I cannot find any reference to J&W 1972 on your site (in fact, not
anywhere)

2/ I am aware of:
The Programming Language Pascal. Acta Informatica, 1, (Jun 1971)

3/ I am aware of:
The Programming Language Pascal. (Nov 1972)

4/ I am aware of:
PASCAL - User Manual and Report. Jensen & Wirth (1974)
Reprinted/edited 1985 & 1991 subsequently gaining the subtitle "ISO
Standard Pascal"

To me the *earliest* document defining the Pascal Language provides the
*original* definition. (I stress the word ORIGINAL - as I have all along.)

To me the biggest difference between the ORIGINAL (as described above)
Pascal and the ISO standard is in the matter of extensions.

The Original Pascal allows for predefined Procedures and Functions but
the ISO standard goes further and allows for redefinitions of chapter 6:
hence your IPPascal may accept the underscore in identifiers (or someone
could have an extension allowing a digit as the first symbol in an
identifier!). Original Pascal did not allow for such variations on the
syntax of the language.

The interesting point is that the standard is very forgiving in what the
PROCESSOR is allowed to accept but at the same time very strict on what
is allowed in a conforming PROGRAM. So if someone writes a program
that can be happily processed by a particular conforming processor, that
PROGRAM does not neccessarily conform to the standard and is therefore
not neccessarily portable.

IMHO a standard should not explicitly allow undefined extensions: I
think that is where a lot of people are likely to become critical of the
standard. If extensions are neccessary they should be defined and
included in the standard, even if it means revising the standard from
time to time. Then any PROGRAM or PROCESSOR could include a statement
advising with which standard(s) it conforms.

scott moore

I can't control your opinions. However, again, I didn't say that.

I'm glad to see you are finding out what is and what is not ISO 7185
standard Pascal.

Scott A. Moore

If that's your main grief, please check the standard again.

5.1 says:

: A processor complying with the requirements of this International
: Standard shall [...]
:
: f) treat each violation that is designated an error in at least one
: of the following ways
:
: 1) there shall be a statement in an accompanying document that the
: error is not reported, and a note referencing each such statement
: shall appear in a separate section of the accompanying document;
:
: 2) the processor shall report the error or the possibility of the
: error during preparation of the program for execution and in the
: event of such a report shall be able to continue further processing
: and shall be able to refuse execution of the program-block;
:
: 3) the processor shall report the error during execution of the
: program; and if an error is reported during execution of the
: program, the processor shall terminate execution; if an error occurs
: within a statement, the execution of that statement shall not be
: completed;
:
: g) be accompanied by a document that separately describes any
: features accepted by the processor that are prohibited or not
: specified in clause 6: such extensions shall be described as being
: `extensions to Pascal as specified by ISO/IEC 7185';
:
: h) be able to process, in a manner similar to that specified for
: errors, any use of any such extension;

So, it is possible for a processor to accept extensions, but then it
must at least document them twice, first as being extensions, and
then as being not reported (or document them once, and at allow to
reject them at compile-time or run-time).

You may thing that's still to liberal, but in reality what more
could you expect? If this wasn't allowed, compiler makers would
probably still make extensions, just they couldn't say "conforms to
ISO 7185 with the following extensions ... and those extensions are
not reported", instead they would say something like "... implements
the language described in ISO 7185 plus the following extensions".
It wouldn't make a practical difference.

BTW, the standard doesn't enforce portability (as you seem to imply
above). Even a perfectly conforming program could be unportable (cf.
5.2 note 2).

The standard requires documentation of implementation-defined
features (here: MaxInt) (5.1 d). So you need to consult the
documentation of the processor and evaluate your program with
respect to it if you want to ensure portability (or check your
program against the standard). This goes for extensions just as well
as for implementation-defined features.

Frank

--
Frank Heckenbach, frank@g-n-u.de, http://fjf.gnu.de/
GnuPG and PGP keys: http://fjf.gnu.de/plan (7977168E)
Pascal code, BP CRT bugfix: http://fjf.gnu.de/programs.html
Free GNU Pascal Compiler: http://www.gnu-pascal.de/

scott moore

The J&W 1972 Report is available here, a .PDF of the original:

http://www.moorecad.com/standardpascal/documents.html You're welcome.

Sure. Of course, by that definition the CDC 6000 implementation, and all
other implementations are not original, since Wirth revised the language
in 1972, as covered by the Report, which you have access to above.

Thats simply not true. The Report, and all other Wirth documents not only
do NOT forbid extensions, but the J&W gives a whole chapter dedicated to
the extensions present in CDC 6000 Pascal !

I do understand that this is a common myth spread concerning Pascal
(mostly by Borland advocates who seem to want to "prove" that other Pascal
implementations are as incompatible with original Pascal as Borland is).

However, telling the same myth over and over does not make it true, just redundant.


The standard allows for extensions, just as Wirth's original Pascal did.
The ISO 7185 standard goes farther by asking that complying processors
implement an option flag that rejects non-standard constructs so that
programmers can create absolutely compliant programs if that is
desired.

You asked for two things, above, so I'll address them separately.

First, for the standard to NOT ALLOW extensions is an absurd requirement.
Wirth hallways intended that extensions be present in the language, and
specifically discussed what should and what should not be in the standard
vs. discussed as a local processor extension. Basically, Wirth was saying
that some features were inherently machine dependent and should NOT
be standardized. This is all available if you read J&W.

Now as for "extensions should be in the standard", Wirth, again,
addressed that issue in J&W, and many times thereafter, that some extensions
or machine dependent features should NOT be standardized. If you read
J&W, the CDC 6000 implementation you will find a list of them (for
one sample machine).

Finally, no language is static, and Wirth never intended the language
to be set in concrete, never to be extended ever again. For your
requirement that "If extensions are necessary they should be defined and
included in the standard" presumes that the makers of the standard
knew the future, and could anticipate all needs forever. The
language Pascal was created by Wirth in 1969. The standard was created
in 1982. We are talking about a language almost 35 YEARS OLD !

Thank you for your interest in the standard.