The sources for Minerva - my (the?) replacement ROM for the Sinclair
QL.

Please note my copyright and the GNU general public license

Also, here (39K) is the generated ROM
image along with its.link map (9K)

Some quick notes on Minerva versions, mt.extop and software
reset.

I think TextPad users might like my Minerva 1.98 SuperBasic syntax rules - put them in your "Program Files" for TextPad, inside the "Samples" folder.

After much confusion, I have settled on what I think is the "best" version of Minerva. I had got up to what I was calling 1.99 on my own machine, but it, and an abortive version of 1.98 that I don't think anyone really got their hands on, both suffered from a bug in (Super) Gold Card that meant their SDATE and ADATE calls didn't work. (If anyone actually has a 1.98, here is a little LRESPR code to cure them.

Updated 10/04/2003

A couple of Delphi based programs.

N.B. If you don't have the file VCL30.DPL
(446K) in your Windows/System directory, you will need it, as these
two programs are *so* much smaller without it built in!

I rather like to watch Channel 4's Countdown. I can usually match
Carol Voorderman's performance at the numbers game, but I always found
it frustrating when we neither could solve a particular game. So I wrote
a program... and rewrote it, and rewrote it. On my P100 it now gets a
solution in about a quarter of a second and two seconds later has all
the (shortest) solutions. Carol has since doubly impressed me, on an
occasion when she came up with the *only* solution, which needed
the full six numbers and was not at all an obvious solution.

Updated 29 May 1999 - New in version 1.1.2:

- Factorisation
- Continued fractions.

I know there are various similar programs out there on the web, but I tried a bunch and didn't like them.

This one make a fair try at being friendly (so I hope).

ECalc is the result of me wanting to have a calculator that was pretty easy to use and could give rather a lot more precision than the usual crowd.

I've always liked reverse polish calculators, since the first such Hewlett Packard ones came out.

In terms of precision, I've allowed ECalc to go up to a million bits (300,000 digits) but...

In terms of speed, ECalc is fine up to a thousand digits, say, but
as I haven't worked out how to do FFT multiplication, so the time taken
evaluating functions goes up with the *cube*
of the number of digits. On my P100, it will do 1000 digits of Pi in 0.6
seconds. That's using its clever in-built algorithm. Using the classic
arc tangent of one, times four, takes 2.31 seconds.

In terms of dynamic range, it will cope with 1e-301029995 to 1e301029995. You can calculate number of electrons in the universe with no trouble.

.A fellow QL user, some time back, mentioned that he was having trouble doing some obscure statistical calculations that, for some reason, needed rather high precision intermediate calculations.

At the time, he was using a program written in QL SuperBasic, which did all its calculation using strings of decimal digits. Needless to say, it wasn't all that fast.

Having caught my interest, I went off and wrote some pretty tight Motorolla 68K assembler to do binary arithmetic on a fixed width stack (up to 256 bytes), wrapped it up with a Forth-like front end written in SuperBasic, and handed it over.

Rather a lot of time has passed since then, but I suddenly got the urge to port it to Delphi. The result bears almost no similarity to the original, and this one is a rewrite of an earlier version that wasn't particularly friendly.

I haven't quite finished developing this one.

I may well add in some loop and conditional programming features (à la Forth).

If anyone out there knows of a way to calculate the gamma function, I'd love to add it in! All the algorithms I've come across so far are horrendously naff.

Having gone through a ridiculous struggle to put together the help files for ECalc, plus wanting to be able to simply maintain a copy of my CV in various formats, I'm part way through developing a Rich Text editor that will a) output cleanly as straightforward HTML and b) provide a comfortable, graphic way to construct help files.

Once again, I know that there are programs available, but they are either fiddly to use or cost money.