Somebody shoot me...

No idea why I started doing this. It just sort of happened. Hmm…

It doesn’t attempt much in the way of recovery. The slightest upset and it’ll either crash the machine or quit. Wondered why opening !Boot would hang things up, until I remembered Loader is in there, and it probably couldn’t cope with an image file. Correct, so I fixed that. Still need to see why it hangs instead of just bombing out. It’s doing some oddball stuff to the Wimp given it’s expecting the pre-multitasking version. ;-)

Oh, and Exit exits immediately unless something has noted there’s unsaved data. The Arthur 0.30 version at least asked you if you wanted to quit. Don’t know what they got rid of that.

It is running at a 640×480 display, but thinks it is MODE 15 (otherwise the sizing of icons and stuff is horribly wrong); it’s supposed to work in MODE 12, 20, and 22 which translates to 15, 21, and 28 in 256 colours (because the Pi doesn’t do 16 colour modes).
If I can be bothered, I might rip out all of that and hardcode whatever works in a 2:2 mode.
Of course, the palette tool does nothing, as its expecting 16 colours. I think I’ll need to boot a real Arthur to see what the colours are supposed to be, the clock just feels wrong…

Still, I’ve managed to get that much running after an hour or so of fiddling.

ROFL

Perhaps you saw my “just sort of happened” earlier today?

http://clive.semmens.org.uk/RISCOS/RandPats.html

I actually published 38 of these things on Feyspook, but by screenshotting them in !Paint and converting to .jpg with ChangeFSI rather than going via .svg since Feyspook doesn’t like .svg. I might put a few more of them on my website later.

Perhaps you saw my “just sort of happened” earlier today?

No, but pretty.

I didn’t want to do anything taxing, so… yeah… I probably should have just spaced out with Netflix. ;-)

Just read your info on the ARM ARM. Odd thing is you don’t mention issues or dates.

The two I have, I recall you asked earlier but I forgot to look…

• Probably issue 1. ISBN 0-13-736299-4, ARM DDI 0100B, issue B. July 1996.

You aren’t credited in that, but there’s enough there that I think I ought to be able to decode all of your initials. ;-)

• Second edition (says so on the front). Easily three times as many pages. ISBN 0-201-737191, ARM DDI 0100E, issue E, June 2000.

Differences?

Programmers model 10 pages vs 34 (!).
ARM instruction set 81 pages vs 116.
Addressing modes¹ 43 pages vs 65.
Thumb instruction set 82 pages vs 105.
Enhanced DSP extension 0 pages vs 34.
System control / MMU 31 pages vs 91 (!).
VFP 0 pages vs 149.

That said, I prefer the smaller ARM ARM for two reasons:

1, The heavier print used is clearer. Especially around the instruction encoding boxes.

2, The greyscale boxes at the page edges are not only extremely clear, they can be seen from the side so make it easy to flip to the right part of the book.
By clear, LDM (2) says under it Addressing mode 4. In the bigger ARM, that’s buried in the instruction description text.
MSR says “Architecture v3 and v4 only”, has this stuck in the textual description as well (though this time it says “Version 3 and above” as more families came along).

The smaller book’s register outline is big and bold and clear.

In terms of textual clarity, on the other hand, the bigger book wins. Not a surprise, it’s easily 3x larger. The description of the PSR in the small book usefully follows the register layout, and has a single page talking about the I and F bits, and the processor modes. The big book has four pages, and remembers to say what N, Z, C, V are for. It looks like the smaller book makes a one-sentence passing reference to these. ;-)

So, big book wins for lucidity. Small book wins for layout and clarity.

¹ This is combined with the ARM instructions on the smaller book.

That reminds me – was any ARM ARM published with the UAL syntax?

I prefer dead trees, so an 80,000 page PDF isn’t so useful for browsing…

was any ARM ARM published with the UAL syntax?

Yup – the one I did in 2007 was. ARM DDI 0406A.

That was the first one with UAL, and I think it was the last one printed on paper – and I’m not sure there were very many printed I’m afraid. Might be hard to get hold of. 2078 pages.

Might be hard to get hold of.

Amazon (France) only has the first and second.

2078 pages.

Hmm, and what would cost more, the book or the postage? ;-)

Just saw this three star review for the first edition:

I guess that person doesn’t know what the word “architecture” means in respect to processors, or the fact that ARM licence cores and all that other stuff gets bolted on by someone else.
< facepalm >

I think I was lying anyway, sorry. I think it was the Assembler Guide I was working on at the same time that was printed, and probably only enough copies to be delivered with the assembler to the customers. I’ve a feeling the ARM ARM wasn’t printed. It was talked about, but iirc it was decided not to go ahead with it.

I retired the moment the beast was approved – I’d only stayed on to get it finished, I’d have retired a few months earlier otherwise.

For what it’s worth, I have the same two editions that you have – the first published before I joined ARM, the second before I was involved with the ARM ARM – I was writing Assembler Guides and other assorted documentation at that time.

And I’ve fixed my recount to mention the edition I worked on, and the date! 8~)

That Arthur desktop reminds me that in the few months while I was saving up for an Archimedes A310 in the summer of 87, I wrote my own Arthur like desktop on the BBC Master. It ran in Mode 1 and used 5×7 font to try to look higher resolution. The windows were movable, resizable and scrollable, although not in exactly the same way as Arthur as I’d only ever seen a screenshot of it in the Acorn User at the time. As it was written in BASIC, it still runs today under my Graphic TaskWIndows.

Nice, Druck. Looks good, given what it was intended to be running on.

Better, I might add, than the “desktop” supplied with the Master Compact Welcome disc. Just tried to get a screenshot of that, but BeebIt hangs when trying to be a Compact (hmmm? I wonder what’s wrong) and the Master 128 isn’t quite specced up enough to run it.

I like that the calculator has hex. That’s a nice touch.

The windows were movable, resizable and scrollable

Written from scratch, in BASIC. Great job!
[Arthur uses a rudimentary incarnation of WindowManager]

That looks genuinely quite amazing, and considerably nicer than the “GEM” desktop supplied with the Master 512. (Which I’ve been playing with this week. It’s quite hard if you don’t have the special mouse.)

It looks better than it works, as it was never finished due to actually getting the Archimedes. But if you want to take a look it’s !DEEJsys in GraphTask’s demo directory at https://armclub.org.uk/free/

BTW: the screenshots seem to have disappeared from this topic.

the screenshots seem to have disappeared from this topic

Never saw them at all, I thought you were all referring to some resource/site discussed elsewhere. So, they were there and now they aren’t.
I blame Rick, he’s got “form”¹ so that’s enough excuse.

¹ Inserting codes for sounds in, was it mail or news?, and annoying various Argonet subscribers.
Of course, I had nothing to do with any of that, honest (maybe)

Strangely, the screenshots have been there from the beginning for me, and still are.

That is, one from Rick in the first post, and one from David at 9:38 pm on the 12th.

Strangely, the screenshots have been there from the beginning for me, and still are.

Some browsers won’t display them, as they are HTTP images while the owning page is HTTPS.

Some browsers won’t display them, as […]

Interesting. I’m using Firefox on two different Macs (a laptop running High Sierra, and a Mini running Mojave). Not checked on Safari on either of those, nor on NetSurf on the Pi.

Some browsers won’t display them, as they are HTTP images while the owning page is HTTPS.

Do we know which are the faulty browsers that fail in such a simple task?

Do we know which are the faulty browsers that fail in such a simple task?

It’s part of the drive to banish HTTP from the Web. Mixed-content pages used to display a “broken padlock”; now they have started to omit the HTTP content altogether.

It’s part of the drive to banish HTTP from the Web.

Sort-of.

Mixed content is problematic, because you’ve gone to the extent of securing the page itself, but then pull in insecure content which can bypass the security. Images are considered “passive”, so most browsers currently include them and warn¹ because the worst that can happen is you see an image that isn’t the one that the author intended. A the same time, they block insecure active content.

The move is slowly towards blocking all content, and over time the settings for blocking passive content – which are currently defaulting to “off” – are likely to switch over so that users have to knowingly enable display of the problematic items.

Sites with no HTTPS are not affected and will display as normal… but their search rankings could suffer over time, IIRC.

¹ Firefox has a yellow triangle in the URL bar for this site, which as I explained a while back when I was told it was due to certificate issues, is actually highlighting the insecure images linked by posts.

I blame Rick,

Gee, thanks.

he’s got “form”

What, making other people’s pictures disappear? They would make me, what, digitally omnipotent?

the screenshots have been there from the beginning for me, and still are.

Same for NetSurf earlier, and Firefox (Android) now.

Mixed content is problematic, because you’ve gone to the extent of securing the page itself,

In as much as one can consider it “secure” when your trust isn’t with the site but with a set of random unknown trust agencies (all that CAdata guff).

And even then, quite possibly for a lot of people simply because they are being bullied into it.

are likely to switch over so that users have to knowingly enable display of the problematic items.

Hmm, warn me if the cert is a dud. Warn me if there’s mixed content. But don’t bloody nanny me. The Facebook tracker icon is https and I consider that more of a risk than Druck’s image, for a number of reasons that have to do with it being accepted by the browser, not whether or not it is https. Again, we’re back to somebody else deciding what is “safe” for us.

Luckily my Android Firefox is the older pre-broken version that has some fairly aggressive content blocking included, so I never see the thumb except here on this site…because cleverly they serve it up themselves rather than dropping in a reference to FB’s site each time. ;-)

Now, to the topic in question. I didn’t get far hacking SDFS into DEEJsys. Probably a combination of long filenames and way more entries than a Master would have supported in any directory on any filing system…

But that aside, it’s actually a bit bloody brilliant. A windowing system, written in a remarkably small amount of code considering, that fluidly supports arbitrary window placement, size, scroll position, and (via outline) movement. Redraws seem reasonably quick. There is a little lag, not sure if that’s BASIC or the via-TaskWindow graphics stuff (sadly it gave a corrupted display when running native as the Pi only supports 256 colours minimum). More impressively it doesn’t flicker like crazy and the redraws are competent, even with a window over a window. There’s no junk left on the screen.

It is, in actual fact, a pretty damn good attempt at replicating the Arthur desktop. That it was intended to run on a Master is made of win.

Speaking of which, when it borks it pops up a nice clear error message with the option to continue or to abort.
That, alone, means it’s just bested Arthur. Because there are many “unexpected conditions” to which the Arthur desktop program will just STOP.

So, all in all 👍.

Do we know which are the faulty browsers that fail in such a simple task?

Android Chrome 84.×. There’s no workaround.
https://blog.chromium.org/2019/10/no-more-mixed-messages-about-https.html

Gee, thanks

It’s a pleasure :)

Note: very recent Firefox upgrades may actually be the cause of people previously seeing and now not seeing those images.