[Simh] Pascal 1.3 manual for RSX 11 4.6 - warning, nostalgia trip!

[Johnny Billquist]

On 2016-02-05 00:36, brian wrote:
> On Tue, 2 Feb 2016 15:03:36 +0000, Jason Armistead wrote:
>
>> Al Kossow wrote:
>>
>>> here is the SPD
>>>
>>> http://h18000.www1.hp.com/info/SP1418/SP1418PF.PDF
>>
>> And the second page of the SPD even mentions "On-line Debugging Technique (ODT)" - the topic of one of our other recent SIMH mailing list threads !!!
>>
>
> It's more than 30 years now since I last used Pascal on the PDP-11
> (RSX-11M 4.1, no PLUS). AFAIK, RSX Pascal was derived from what we
> had, which was Oregon Pascal. I believe that DEC subsequently bought
> it to release as the first version of RSX Pascal. Oregon Pascal was
> one of two Pascal compilers that were available for the PDP-11 in
> those days, I think the other one was called Whitesmiths, but I'm not
> 100% sure.

I don't think that the DEC RSX PASCAL compiler is not related to the 
Oregon PASCAL compiler.
I have both around, but I never use the DEC one, but I do use the Oregon 
PASCAL from time to time.

There was at least one more, which was the "Swedish Pascal", which 
were/is available from DECUS. It's pretty okay, but there are some ugly 
code in it which makes it hard to adapt to a more modern world.

> Oregon Pascal was BY FAR superior, we found out after a trial run or
> two that Whitesmiths Pascal was actually a translator rather than a
> compiler, they had a 'C' compiler, and their Pascal 'compiler'
> actually translated to 'C' and then compiled the resulting 'C' code.
> If you told it to stop at MACRO-11 sources (both compilers could do
> this) then the Whitesmiths code was MONUMENTALLY inefficient compared
> to the Oregon code.

Whitesmith definitely had a C compiler. I think I even have it 
somewhere, but I have never tried it. But the stories I've heard have 
not been favorable. I wouldn't be surprised if they did PASCAL as just a 
frontend to C, but it would be ugly I'm sure.

> I missed the thread about ODT, but I sure remeber having to use it,
> The debugger supplied with the compiler was just about capable of
> debugging 'Hello world' before the code got so large that it broke the
> limits, even with overlays. All of our debugging had to be done by
> compiling to MACRO and then patching breakpoints into the executable.

I don't think I've seen anything about any other debugger provided with 
RSX Pascal, so I would assume that you only had ODT to choose from. Are 
you saying that there was some other debugger provided as well? That 
could be interesting to look at. Oregon PASCAL certainly do not have 
any. (Unless my memory is playing tricks with me, I have the manual 
somewhere...)

> The other 'interesting' part of it was writing the overlay structure
> for the linker. For a big program, trying to get every branch within
> 32 KB  could be almost as much work as writing the code! The hell with
> logical program design, you wrote your code first and then came the
> tricky part, hacking it all around to get within that limit. I
> remember breaking the linker (task builder) at one point because of
> trying to break the code down into too many branches.

Ah, yes. Working with overlays can be a lot of "fun". A really good 
reason to move the RSX-11M-PLUS and capable hardware, as you then can 
use split I/D space as well as supervisor mode libraries, which helps 
tremendously with memory usage. Many times you can just skip overlays 
thanks to that.

> I didn't see what DEC did to the libraries after they bought Oregon
> Pascal, but I am 100% certain that all the libraries we had to link
> against lived under [1,1]. I was (I think) the only programmer on the
> team who used DCL in preference to MCR, but our editors were K52 (for
> the VT 52 terminals) and KED (same editor but for the VT 100s). There
> was also a MAKE utility which was available, but the problem with that
> was that it burned so much CPU trying to work out what to compile on a
> big program that the PDP-11 just slowed to a crawl (there were about a
> dozen of us developing on that one machine) so it ended up that we
> were only allowed to run one MAKE a day after everyone had gone home.

Yes, LB:[1,1] is where pretty much all normally used libraries gets 
installed. Not required, but a convention.
That is equally true for DEC software as well as other third party software.

And you must have had a pretty sucky make system. :-)

> Fun times, though. :)

Indeed.

	Johnny

-- 
Johnny Billquist                  || "I'm on a bus
                                   ||  on a psychedelic trip
email: bqt at softjar.se             ||  Reading murder books
pdp is alive!                     ||  tryin' to stay hip" - B. Idol