Noone seem to have brought this up, so I guess I'll have to.
Standard Pascal is pretty straight forward, and usually compiles easily
from one compiler to the next.
However, Pascal was notorious for omitting how to do I/O to anything but
the terminal. The standard does not cover this, and every compiler had
to come up with some extension for it. And thus, of course any code that
does anything related to files, will, by definition, be non-portable.
In this case, you have the assign() function which is used to associate
actual filenames with handles inside Pascal. In VMS Pascal, you do not
use a function called assign() for this. Check the VMS Pascal manual,
but I suspect it might be called open() maybe? And of course have
different order and types of arguments, so you cannot just change the name.
File I/O was *always* the first thing brought up whenever talking about
the limitations and shortcomings of the Pascal standard.
Johnny
On 2018-02-08 18:21, Dan Gahlinger wrote:
so here you go, a simple file compare I wrote using "freepascal"
(freepascal.org) and I've done what I can to convert it to vms pascal
but it doesn't compile.
code:
Program fcomp(input,output);
var
f, g, h : text;
s, t, u : varying [255] of char;
c, d : char;
i : integer;
begin
s := 'first.txt';
t := 'second.txt';
u := 'output.txt';
i := 0;
assign(f,s);
reset(f);
assign(g,t);
reset(g);
assign(h,u);
rewrite(h);
while (not(eof(f))) do
begin
read(f,c);
read(g,d);
i := i + 1;
if (c <> d) then
writeln(h,i,' = 1:[',c,']/',ord(c),' 2:[',d,']/',ord(d));
end;
close(f);
close(g);
close(h);
end.
errors:
pas fcomp.pas
00016 0 1 assign(f,s);
1
%PASCAL-E-UNDECLID, (1) Undeclared identifier ASSIGN
at line number 16 in file DUA1:[DAN]FCOMP.PAS;4
%PASCAL-E-ENDDIAGS, PASCAL completed with 1 diagnostic
------------------------------------------------------------------------
*From:* Clem Cole <cl...@ccc.com>
*Sent:* February 8, 2018 10:58 AM
*To:* Dan Gahlinger
*Cc:* Gary Lee Phillips; Tim Shoppa; simh@trailing-edge.com
*Subject:* Re: [Simh] anyone know how to convert/translate turbo pascal
to vax pascal?
Dan,
As others have said something smells wrong here. It's true the original
'71 report from Wirth did not defined I/O and '72 revised report only
defined write. By the time of the Jensen & Wirth book from
Springer-Verlag in the mid-late '70s writeln is there. And by the time
of the first standard efforts @ IEEE and ANSI it very much in the
language. I do have an old copy ANSI/IEEE770X3.97-1983 "American
National Standard Pascal Computer Programming Language" which on page 93
(Section 6.9.3) defines the required standard Pascal function writeln:
*6.9.4 The Procedure Writeln.*
The syntax of the parameter list of writeln shall be:
writeln-parameter-list = [ "(" ( file-variable |
write-parameter )
| "," write-parameter | ")" ] .
*Writeln*shall only be applied to textfiles. If the
file-variable or the writeln-parameter-list is omitted, the
procedure shall be applied to the required textfile output.
From a quick search on the HP web site
(http://h41379.www4.hpe.com/commercial/pascal/pascal_index.html )I found
reference to the SPD and the site says: :
HP Pascal (formerly known as Compaq Pascal and DEC Pascal) runs on
OpenVMS for VAX systems, OpenVMS for AlphaServer systems, and
OpenVMS for Integrity servers. With HP Pascal, your source code
investment is not only protected, it is extended.
HP Pascal supports code compatible with either level of the ISO
specification, meets Federal Information Processing Standard
Publications (FIPS-109) requirements, and supports many features
from the Extended Pascal Standard. HP Pascal has a solid reputation
as a robust, production-quality, high-performance compiler. It is a
full compiler, not an interpretive one. Tightly integrated wit
While I do not have FIPS 109 on my system, FIPS was based on
ANSI/IEEE770X3.97, and I do have a copy of the an old DEC pascal manual
so I'm 100% sure writeln is there. I also believe that Turbo Pascal was
developed after the ANSI/IEEE770X3.97 was published so the Turbo
extension to writeln and any VMS ones should be able to puzzled out.
Here is a pointer to Pascal for OpenVMS - User Manual Order Number:
AA-PXSND-TK
<https://support.hpe.com/hpsc/doc/public/display?docId=emr_na-c04619822> This
may give you hints.
That said, I have a PDF of the VMS Pascal Reference, but lord knows
where it came from; probably my time at DEC. I have to believe its on
bitsavers or the like. But when I look on page 9-67, Section 9.8.26
defines the WRITELN procedure as defined in the standard with the one
DEC extension of supporting that last optional parameter to be: [,ERROR
:= error-recovery] where error-recovery is defined as the action to be
taken when an error occurs.
As other have said maybe its something silly from file format
conversion like <CR><LF> processing as VMS record oriented I/O is
different than DOS/Windows. But I suspect you are running into a
difference in how I/O is declared and bound to files on the disk. My
experience with a number of different Pascal compilers 'back in the day'
was this was an area for wide variation. This is what I would look up
in the HP/Compaq/DEC user manuals I just pointed you too. I suspect
that the 'VMS Pascal Language' manual should help you through the DEC
variant, so google is your friend to try find a PDF. With that open and
a Turbo Pascal manual I think you'll be fine (and if you cannot find a
Turbo manual, I have to believe the freepascal.org
<http://freepascal.org> docs will get you a long way since they claim to
be 100% Turbo Pascal and Delphi compatible).
BTW: One other though/place where Pascal I/O can differ is character
sets, although I don't think it a problem because PCs and Vaxen never
had this issue, if you read any the reports or "Jensen and Wirth" you
will notice that Pascal was defined for a 6-bit byte on a CDC-6600
system (SCOPE was the OS IIRC). As I have said elsewhere, a 6-bit
character was not unusual on earlier systems -> today, we can thank Fred
Brooks for the 8-bit byte (Gene Amdahl wanted it to be 6 bits but Brooks
kicked him out of his office until he had something that could easily be
handled by SW - /i.e/. a power of 2, which Amdahl thought was wasteful
for the HW).
Anyway, characters can be an issue when moving Pascal code because
original Pascal was defined with some CDC isms and in those days, CDC
had as number of different character sets. I note that if look at the
ANSI standard you'll noted the definition of all identifiers is just the
lower case [english] chars a-zfor letters, the traditional digits 0-9and
very limited number of special-symbols ( + - * / = < > [ ] . , : ' | ( )
). They do say:
The representation of any letter (upper-case or lower-case,
differences of font, /etc/.) occurring anywhere outside of a
character-string (see 6.1.7) shall be insignificant in that
occurrence to the meaning of the program.
Because of the '6-bit ness' of some systems, the standard even provides
for alternative tokens to do things like square braces to (.and .)or
vertical bar to @[and I think may allow ^to be used for same IIRC].
Also remember that DEC manuals tended to show the identifiers in upper
case (go figure). And since both Vaxen and Intel processors (/i.e. /VMS
Pascal and Turbo Pascal) support at least 7-bit ASCII fitting into an 8
bit character, and DEC added support for other special symbols such as
dollar $, but I have to believe the problem is not in character set.
Best wishes,
Clem
ᐧ
_______________________________________________
Simh mailing list
Simh@trailing-edge.com
http://mailman.trailing-edge.com/mailman/listinfo/simh
--
Johnny Billquist || "I'm on a bus
|| on a psychedelic trip
email: b...@softjar.se || Reading murder books
pdp is alive! || tryin' to stay hip" - B. Idol
_______________________________________________
Simh mailing list
Simh@trailing-edge.com
http://mailman.trailing-edge.com/mailman/listinfo/simh
