Programming languages

Precisely that, Steve (F)

Where do I find out about the Software Preservation Project?

On the subject of programming languages, it was officially announced today (3 Dec) that Lua 5.4.2 has been fixed as the current release of Lua 5.4. RiscLua (at least the VFP version) tracks this with an update, version 84. Let me know if you find any bugs or oversights.

I thought that was the point of the Software Preservation Project being discussed at the London Show? And that they were already on to the archive storage (which wasn’t publicly accessible, precisely because of the IPR point)?

The SPP was an item I missed.

Could you remind everyone who also missed it what the link is for the currently available software?

Last year, maybe 2018 even, I did some testing with Tim Rowledge for an updated RISCOS version of !Squeak
4.4, though it can go to 4.6 via the built in updater, after we in MUG arranged for an update to his DDE to be made available to him.

This meant that Scratch 1.4 would work as well on ARMX6 etc.

Where do I find out about the Software Preservation Project?

It is an ABug project, following up their preservation efforts regarding the 8bit Acorn stuff. Phil Pemberton did the presentation during the virtual London Show.

Some videos about the ABug preservation: http://abug.org.uk/index.php/category/preservation/

It is an ABug project, following up their preservation efforts regarding the 8bit Acorn stuff. Phil Pemberton did the presentation during the virtual London Show.

OK, I thought there was some separate effort. I know who to talk to about archiving various kinds of software (games, education) but don’t remember details of any general effort. I think there was a follow-up meeting about setting up a site for preservation. I guess I need to ask on stardot to see what the current status is.

As it happens I’m currently doing a small update of Scratch 1.4 for RPF and at least in principle it should transport to RISC OS successfully.

What I don’t have very much hope of finding time for is updating the virtual machine to the Cog version. Although i’d bet that the actual code generator would work with no problems – it isn’t OS dependent at all, other than handling ‘fun’ with write execute permission issues in some systems – there are a number of things that might cost time. One issue is having a decent thread capability to support an internal ‘tick’ that drives some important routines. All the support from bcc intrinsics would need to be replicated as well,

So, hope for some improvements but probably not all. Unless of course, someone hands over…
One. Meeeeeeellion. Dollars!

@ Tim

So, hope for some improvements but probably not all. Unless of course, someone hands over… One. Meeeeeeellion. Dollars!

In which case may I join you in the coding effort? :D

Gavin: The link in your post to the RiscLua84.zip file is broken; you’ve linked to it as //http:www.wra1th.plus.com/zip/rlua84.zip – i.e. your finger that hit the slash key twice was so quick off the mark it beat the fingers that typed the http: ;)

The link in your post to the RiscLua84.zip file is broken;

Whoops. Sorry about that. It is now corrected.

Paolo, should I ever get that million I shall endeavour to send you a couple of bucks, just for the sheer cheek of it.
David, don’t forget that the ‘t’ is lower case – this is totally crucial.

There’s actually quite a few modern Smalltalk systems out there, including Squeak and it’s forks (Cuis and Pharo) and VisualWorks (which I used to be departmental manager for many moons and a country ago), the (originally IBM & partly started by me) VA Smalltalk, a descendant of Visual Smalltalk Express, Dolphin, MT, Smalltalk/X, fScript and probably a bunch of others.

The interesting oddball is SqueakJS, which as the name implies is Squeak on JavaScript. This turns out to be a quite clever thing, though it does really need a quite fast machine to be interesting. It has, amongst other things , allowed the creation of The Smalltalk Zoo, a Computer History Museum ‘exhibit’. See https://computerhistory.org/blog/introducing-the-smalltalk-zoo-48-years-of-smalltalk-history-at-chm/ for an interesting learning experience.

There has been Smalltalk available on ARM since before ARM was available, if you see what I mean. I hope we can keep it that way for another few decades at least.

It amazes me that afaict there never was an APL system for RISC OS. With its graphics and font capability I would think APL would be a natural fit?

There is a J (26 bit, no source) but it’s just not the same…

It amazes me that afaict there never was an APL system for RISC OS.

I have one, but it is 26-bit, as is the J I have. I also have quite a run of the newsletters somewhere.

APL: the ultimate write once and never read language. I was fascinated by it, in the same way a rabbit is fascinated by an adder. I first had it on the BBC and then the Arch. I tried examples and even adapted some, but I could never get my mind around it. ;-)

APL: the ultimate write once and never read language. I was fascinated by it, in the same way a rabbit is fascinated by an adder.

Not being a programming language aficionado APL was unfamiliar.
I’ve just done a Google query and the first phrase that came to mind was “obfuscated Perl, in Russian”

To quote E.W.Dijkstra: APL is a mistake, carried through to perfection. It is the language of the future for the programming techniques of the past: it creates a new generation of coding bums.

Perl is line noise to manipulate text, APL is line noise in a High Math charset for manipulating numbers. :-)

There is some truth to the “problems of yesterday” quote. I played with APL a fair bit, but just didn’t need to push huge vectors of numbers around…

Other languages (with source for a 32bit conversion) I found on ADPL:

BCPL. Has some weird assembly bits, I’m confused as I’m a little familiar with the ancient CINT version.

Icon v8. ONE assembly file away from a possible recompile. I liked Icon, but I think python killed it.

sbprolog, huprolog, gawk 2.11. Pure C.

Other 26 bit langs I have off the Net:
siod. Another scheme. pure C.
imp77. mutant algol from Edinbourgh. has source, in IMP77… Also seens to have some sort of Pascal compiler if you’re that desperate for Pascal…
pot. Oberon2C but only has a bare-bones runtime.
ptc. Pascal2C This one is weird, wants to patch toansi and I just dunno man.
Xlisp+ 2.1 pure C, 32 bit recompile is on my web site
Xscheme 0.28 ditto

APL: the ultimate write once and never read language.

Ugh. Gives me the shivers just to remember it – including the special APL keyboards with an extra hieroglyphic on eack key. One line of APL code could do amazing things quickly … but normal humans took for ever to understand it.

XLisp and XScheme were created by David Betz, who also created Bob. He was an editor of Dr.Dobb’s Journal. I had this email from him on 02 December 1994:

Your modification of Bob for ARM machines sounds very interesting. Go right ahead and distribute it any way you feel is appropriate. Just keep my copyright notice attached (along with you own if you want). Also, could you send me a copy? Unfortunately, I won’t be able to run it because I don’t have an ACORN machine, but I’d like to see your documentation and what you’ve done with the source code. Are you planning on releasing your source as well? Just out of curiosity, how expensive are the ACORN machines?
Do you know if they are available in the US?
David Betz

ArmBob (feeble name, I am afraid) ran a bit slower than BASIC, but was quite good with wimp programs. I gave up developping it when Lua came along, because Lua was faster and more capable in lots of ways. But ArmBob probably presented fewer challenges to BASIC programmers. I still have the sources if anybody wants to play with it. The C sources are exceptionally well written.

I’m assuming that Willard means that it’s plain old C with no assembly, no BASIC, no POSIX extensions, etc.

Heh yeah, sorry, I was using “pure C” to mean “C with no assembly or other weird 26bit-only components”.

I want to learn ARM assembly, I just haven’t had the spoons (as they say these days.)

Building 32bit xlisp and xscheme, for example, modulo a wierdness with color handling (colortrans involved itself), was compile-and-done. And yes, David Betz’s source is a joy to read. That dude can code! So can Gunnar Zoetl, who did the original conversions and added the WIMP frontend.

Building a new BCPL? That’s another story…

I want to learn ARM assembly

Which version(s)? v8 is very different from v7 and earlier versions. v7 was largely (but not entirely, in important ways) a superset of earlier versions; but v8 is a completely different beast really.

Indeed so, but makes use of very little of the power of later processors.

Oh, indeed. But then since I wrote the instruction set chapters of the v7 ARM ARM I would think that. But of course those ops won’t work on the old 26-bit architecture.