[Expo-tech] 1984 Austria survey data & AERW's software

[Andy Waddington on Loser]

Sometime before sending "Philip Sargent \(Gmail\)" <philip.sargent at gmail.com> typed (and on Fri, 15 May 2020 03:34:41 +0100 sent):

> A weird mixture of Unix code that you could run today and
> very, very obsolete 

Unix ? I don't think I'd ever used Unix by 1984

> Phoenix job control language for the IBM 3081,

Well, it would have been the 370 at the time. 3081's are a bit
modern... by 1984 it would not have been running anywhere near
Cambridge, and would have been on JCL, but still on punched
cards. Unless it was being run in Cambridge by someone else...

> also documents in EGCAL format and Andy's Fortran code (which
> I'll bet runs just fine).

That would be original Fortran, what we learnt in sixth form,
not this new-fangled modern Fortran 70 stuff...

There are versions in other languages too, though possibly only
on paper. I remember rewriting it in APL, just for fun because
it looked so wierd... There definitely wasn't a version in Cobol,
though :-)

> There are no dates on any of the survey files except for cryptic notes
> 
> such as "Planc/AERW 1982 good compass".

There _ought_ to be a date for that in the logbook. "good compass"
means not the compass we calibrated and decided was 134° out (so-
called "fridge north"). Later experience suggests that we calibrated
it whilst standing on Fritz's jetty (mostly made of tubular steel)
and sighting on the Trisselwand, so it is perhaps not entirely
surprising that we got a daft result, and it almost certainly wasn't
because the compass had been left on top of Julian's fridge for
several months. I believe there exists some doubt as to the
sobriety of the calibrators, too...

I'm willing to bet those data are now (well, five years ago - my most
recent copy) stoger.svx, dated 1982-08-02 which says
"ref.:	back of 1982 logbook, and data from AW's coding sheets
 (for punched cards)"

Hmm, those old surface survey files don't half contain a lot
of years-after-the-fact blunder analysis and sketchy fixes :-(

> PS Mostly ascii but the odd bizarre character which must be
> EBCDIC for [ and ].

Yes, the program ran on the University IBM 370 mainframe, from
punched cards, and later on an IBM 360 at Windscale (though it
must have got shifted across to the ICL 2980 at Risley at some
stage, because they had a big drum plotter which we didn't. I
clearly recall getting a rolled up plot back (by UKAEA internal
mail from Warrington), bearing the hand-scrawled comment "your
graph is screwed" or something similar. Not the case - that one
was a perfectly kosher centre-line of Stellerweg :-) I've
almost certainly still got it.

>         4.12   ROSE            rose diagram routine
> 
>         4.13   ROSPLT          package dependent graphics

Those bits really were quite handy for giving clues about the
direction of faults and joints controlling passage direction.
I don't think survex does this sort of thing any more ? A bit
approximate, as legs don't go along the centre of passages,
but zig-zag from wall to wall.

After 1984 the whole lot got ported to a BBC micro in BBC Basic
with some assembler which was needed to load overlays (couldn't
fit all the code in memory at once and also process as many as
1000 survey legs without overlays...). I've probably still got
that, on 5¼" floppies which are not a lot of use...

Then Shawn wrote a whole new program that did loop closure for
these dreadful new IBM PC things (before anything resembling a
proper operating system was written, by a nice guy from Finland)
which had the advantage of 640k of memory (no-one would ever need
more than that, said Bill Gates...). I think that was in Pascal ?

Andy