New Dedicated hardware For RO64 - speculations

Perhaps Peter Pan is good at clearing drains :-/

RISC OS has beautiful WYSIWYG fonts, but missed out on screen color correction for CMYK printing. That made me blue.

The font rendering on RISC OS was, it has to be said, its best feature. Having said that, other systems have caught up and implemented the same type of anti-aliased display. MacOS does it, even Windows does it now. (In fact Windows has done it since XP, it was just turned off by default.)

Colour correction? I’m typing this on a Windows 11 machine with a colour calibrated display. The calibration tools are built in to Windows itself (and have been since Windows 7, possibly even Vista although I never seriously used that). The Mac has always had colour calibration.

If I need it to be completely 100% accurate for proofing purposes (rather than merely ‘indistinguishable by eye’) then I can get hold of something like a Datacolour Spyder and generate an exact colour profile for each of the two monitors I’m using (both Acer, both IPS, but one’s built in to the all-in-one PC and the other is a matching one that uses the same type of panel).

In reply to Rick…

…anybody who in the mid ’20s wishes to implement a shiny new operating system espousing the best of mid ’80s state of the art…
…is a complete idiot.

Sounds to me like you’re having a bad week. Understandable.

Trust me, it’s like arguing with a current-wave audiophile (audiophool?) who is convinced that “vinyl sounds better” and when asked why responds with “because it’s analogue”. Or tries to convince you that PC-OCC directional Ethernet and USB cables will improve the sound of your music streamer? Or the best one I’ve heard – that a TOSlink (optical) cable with gold plated plugs will ‘give a better contact’ and ‘improve the sound’.

(Just let that last example sink in for a moment?)

If you want to make a shiny new OS, you wouldn’t start from here.

I’ve said this before. If you want a modern “RISC OS” for the 21st century, write an API veneer and a work-alike for the window manager, then drop it onto Raspian.

Just let that last example sink in for a moment?

The last one?
I’m still trying to stop my mind rebooting when it hits the phrase “directional Ethernet cable”

Or the best one I’ve heard – that a TOSlink (optical) cable with gold plated plugs will ‘give a better contact’ and ‘improve the sound’.

You’re not buying the idea that a single atom thick layer on the end face will selectively filter out noise, then? :)

TOSLink: I’m still not clear why they had to go off and design a totally different connector than network guys use, and then give it zero rotational symmetry, thus making it equally difficult for the average consumer.

I’m still not clear why they had to go off and design a totally different connector…

I don’t know about this particular case, but the usual reason is “We want to sell more stuff” isn’t it?

There’s a reason every other operating system on earth is arguably slower than RISC OS (on equivalent hardware).

Linux is faster than RISC OS on the same hardware, as it can use multiple cores, where as RISC OS does everything on one. Even Linux single thread programs run faster with time slicing than RISC OS single tasking sharing a core with the OS.

You could improve RISC OS perforamance while sicking with single core co-op MT, by offloading general OS, networking and graphics on to the other cores. But it’s a lot of work for a small gain.

The font rendering on RISC OS was, it has to be said, its best feature. Having said that, other systems have caught up and implemented the same type of anti-aliased display.

RISC OS fonts are nice, but they are stuck in the age of CRTs. Other systems caught up and went further taking advantage of sub pixel anti-aliasing on LCDs, giving close to 3x the apparent horizontal resolution, which is a real advantage at small point sizes.

This is something which could easily be implemented for RISC OS, and would again give us the best looking fonts on LCDs instead of the current rather fuzzy look.

You could improve RISC OS perforamance while sicking with single core co-op MT, by offloading general OS, networking and graphics on to the other cores. But it’s a lot of work for a small gain.

Has anyone done any tests that measure the effect of a sustained network transfer on other processes?
Or tests on a sustained disc to disc transfer on other processes?

Do any such test utilities exist?

TOSLink: I’m still not clear why they had to go off and design a totally different connector than network guys use

Wikipedia suggests that TOSLink dates from 1983. Doesn’t that predate optical networking?

Steve P:

I’m still trying to stop my mind rebooting when it hits the phrase “directional Ethernet cable”

It’s an actual thing:

https://www.futureshop.co.uk/audioquest-diamond-rj-e-ethernet-cable

If that doesn’t make your mind reboot then nothing will!

David:

Other systems caught up and went further taking advantage of sub pixel anti-aliasing on LCDs, giving close to 3x the apparent horizontal resolution, which is a real advantage at small point sizes.

I thought RISC OS had sub-pixel anti-aliasing? Just it was off by default.

It has to be said though, Times Roman on Windows (even with Cleartype) is a lot less readable than Trinity at the same point size on RISC OS. Whether that’s down to the font, or that Windows uses a 72dpi metric rather than 90dpi (as used on RISC OS and on the Mac) is anyone’s guess. I do know that when I view a web site in Windows with the font set to Times I’m straight away thinking “this is hard to read on-screen, why didn’t you set Verdana in the style sheet?”

This is something which could easily be implemented for RISC OS, and would again give us the best looking fonts on LCDs instead of the current rather fuzzy look.

Or use a 4K monitor with the eig set to 0? (I’m assuming RO on the Pi supports that.)

[…] fuzzy look.

I’m surprised. The same text on the same monitor with RISC OS looks much sharper than with Windows 11. Words like milli are fuzzy on Windows and even show a blue-ish flush, but are crispy as any other text on RISC OS. It seems that RISC OS’s sub-pixel anti-aliasing (which exixts as long as I remember) works better than the Windows one.

Wikipedia suggests that TOSLink dates from 1983. Doesn’t that predate optical networking?

“In April 1977, General Telephone and Electronics tested and deployed the world’s first live telephone traffic through a fiber-optic system running at 6 Mbps, in Long Beach, California. They were soon followed by Bell in May 1977, with an optical telephone communication system installed in the downtown Chicago area, covering a distance of 1.5 miles (2.4 kilometers). Each optical-fiber pair carried the equivalent of 672 voice channels and was equivalent to a DS3 circuit. Today more than 80 percent of the world’s long-distance voice and data traffic is carried over optical-fiber cables.”

NB. Quoted attenuation on TOSLink is way higher than even the first network fibre links. Then we modernise.
In Y2K standards: Max TOSLink length 10m @ 125Mb/s?, bog-standard OM1 275 metres @ 1Gb/s, OM2 550metres @1Gb/s
I’m sort of suspecting that if you ran a household audio optical link over a fairly standard bargain-basement OM1 standard patch cable, the performance would be better with the limiting factor being the transmitter/receiver equipment pair.

as it can use multiple cores

That’s not exactly equivalent then. :-)

But, yeah, bring in GPU acceleration and those other cores and, well…

fonts on LCDs instead of the current rather fuzzy look.

It’s worth pointing out that modern anti-aliasing (such as ClearType on Windows) will offer various samples to pick from as it is trying to determine your LCD arrangement (like RGB or BGR) in order to provide the most appropriate output for your display. RISC OS takes no notice of all of this as it is from an era when the dots were phosphors on a cathode tube and there wasn’t actually an exact 1:1 correspondence between screen pixels and actual dots on the screen.
After all, the three electron guns are in a triangle so while they are usually well aligned, it isn’t perfect, so there’s some inherent fuzziness. It made good output on CRTs, but it suffers when faced with the pixel-exactness of LCDs.
Ultimately, yes, I find FontManager’s output to be slightly fuzzy.

Words like milli are fuzzy on Windows and even show a blue-ish flush

I’m not sure if the bluing is because Windows’ anti-aliasing is a bit crap, or because you have it set up wrongly (if you can find where to tune it, do so).
I have noticed that Windows XP did sometimes pick some peculiar colours when scaling text down. I can’t comment on the current as I never progressed beyond XP (saw no pressing reason to).

My Android portable, to a 10" 1024×800 screen, is showing perfectly sharp text at various sizes including small. Visually, it is better output than RISC OS.

I’m still trying to stop my mind rebooting when it hits the phrase “directional Ethernet cable”

Must be the sort that they use for training AI bots. Garbage in…

Steve P:

I’m sort of suspecting that if you ran a household audio optical link over a fairly standard bargain-basement OM1 standard patch cable, the performance would be better with the limiting factor being the transmitter/receiver equipment pair.

A friend once bodged up a link between his CD player’s co-ax output and his minidisc player’s SPDIF input using the horrible cheap 10p plastic phono plugs that Maplin used to sell, and some 13-strand zip wire.

It worked perfectly. Right up until one of his housemates turned on the fluorescent light in the kitchen below his room. Then it didn’t. At all. (Until the light was turned off.)

See, that’s the thing with digital links that audiophools just don’t understand.

Rick:

I’m not sure if the bluing is because Windows’ anti-aliasing is a bit crap, or because you have it set up wrongly (if you can find where to tune it, do so).

Usually the ‘fringing’ on Cleartype is caused by connecting a TFT monitor up using a VGA (D-sub) connector rather than DVI or HDMI. This method requires converting the frame buffer to analogue VGA, then re-digitising it in the monitor.

As an aside, the number of times I’ve gone on-site to find someone has a high-spec PC and monitor connected together with a VGA cable, leaving the DVI ports empty on both the monitor and graphics card. And they wonder why the picture is crap? “That’s the cable it came with though?” Yeah, and shame on PC vendors for not supplying a DVI cable (or more recently HDMI).

My current desktop (all-in-one Win11 i3) supports a second monitor but only has HDMI out, no D-sub. That’s the way things are going, and about time too.

Or then there was the time I went round to a friend’s who was paying a fortune for Sky+HD, had a really decent (for the time) 50" full HD plasma telly… and had the Sky HD box connected to the telly via a SCART cable. Apparently that’s how the Sky installer connected it up. She mentioned that she had a “weird looking cable that came with the Sky box” in the drawer – which turned out to be an HDMI cable. I changed them over and she was fairly astounded at the improvement.

I’ll say it again, this was a Sky engineer that had connected her HDTV to her shiny new Sky box using a SCART cable. Again, let that one sink in.

My Android portable, to a 10" 1024×800 screen, is showing perfectly sharp text at various sizes including small. Visually, it is better output than RISC OS.

My current phone (Motorola g22) has a 720×1600 pixel screen. Given the screen size, that’s about 300ppi. So basically the same resolution as print. Text is pin-sharp even with italics in small point sizes. And that’s an el-cheapo phone. I’ve ordered a g54 as an upgrade, which has a 1080×2400 resolution on the same size screen. That’s nearly 400ppi.

Ultimately, yes, I find FontManager’s output to be slightly fuzzy.

I mentioned in an earlier posting, but I would love to see RO running on a 4K monitor with 180dpi settings (xeig / yeig set to 0). I wouldn’t imagine it would be fuzzy then.

But bearing in mind that RO Font Manager uses a fairly simple AA algorithm. It plots to a ‘virtual’ screen that’s double the size (both X and Y dimensions), where one physical pixel maps onto four logical ones, then it simply counts the number of logical pixels that are turned on and sets the shading of the physical pixel accordingly. (Ok, that’s quite a simplified description but you get the idea).

I’ll say it again, this was a Sky engineer that had connected her HDTV to her shiny new Sky box using a SCART cable. Again, let that one sink in.

I, too, have seen this at a friend’s house. To top it off, her ‘knowledgeable’ son informed me that everyone knew this was the best way to do it.

See, that’s the thing with digital links that audiophools just don’t understand.

Experience says that most electricians don’t understand why data and power need distance between them, and also that data wiring is more than simple electrical continuity for some “low voltage stuff”

I’ll say it again, this was a Sky engineer employee that had connected her HDTV to her shiny new Sky box using a SCART cable.

Fixed.

I mentioned in an earlier posting, but I would love to see RO running on a 4K monitor with 180dpi settings (xeig / yeig set to 0). I wouldn’t imagine it would be fuzzy then.

You’ll have to visit Greenock then. I’d send you a screenshot, but… 8~)

Stuart:

I, too, have seen this at a friend’s house. To top it off, her ‘knowledgeable’ son informed me that everyone knew this was the best way to do it.

Back in the days of standard def Sky Digital, yes, it absolutely was. Use a fully-wired SCART cable and set the output of the box to RGB component. When you had a standard def digibox or DVD player, and a CRT TV. He would have been right – a few years earlier.

Mind you, I have issues of What Hi-Fi from the late 1990s (when DVD players first started becoming a ‘thing’) claiming that you should use an S-Video connection in preference to RGB SCART. Apparently this gave the best picture. (It didn’t – RGB SCART did.)

Also if you connected a Sky box via S-video (I think it was only Sky+ that had this) the TV wouldn’t automatically switch between 4:3 and 16:9 mode. The box wouldn’t generate a Line 23 WSS signal, and required the relevant pin on the SCART cable for the TV to automatically switch aspect ratio.

I think they fixed this with a firmware update on the later Sky+ boxes, but by that time everything was moving to HD anyway.

Steve P:

Fixed.

I’m going to stick with ‘engineer’, in a very loose sense. He’d installed the dish and run the cables. Installing a dish involves measuring, drilling and aligning, which are sort-of engineering type skills. But you make a good point. Maybe “installer-droid” would be a better term?

Installing a dish involves measuring, drilling and aligning, which are sort-of engineering type skills.

I’d call those technician skills rather than engineer skills. Technicians are often better at that kind of thing than engineers, whereas designing stuff is engineers’ work.

I’d call those technician skills rather than engineer skills. Technicians are often better at that kind of thing than engineers, whereas designing stuff is engineers’ work.

Largely speaking, it’s the difference between following a set of instructions that include an S.o.T element and creating the whole package with a set of instructions to follow.

I think the term is “Sky installer” by most accounts I’ve read.

A bit like “Currys expert” then? They started off life as a bicycle retailer – technical skills haven’t improved much since.

Well to be honest I wouldn’t have a clue how to install and align a satellite dish. Bolt this bit to the wall, attach the dish, then do some witchcraft with a compass and a satellite finder.

Once the downleads are into the house it’s easy. F connectors to the back of the receiver (or if you’re being really neat, bring them out to a twin F socket wall plate), along with power, Ethernet and HDMI output to AV receiver.

Again, I’m reminded of a friend who moved house, reconnected his home cinema system by plugging everything into the TV (games console, blu-ray player and Virgin HD box) rather than into the AV receiver. He did manage to get sound out of it as he’d connected the output of the AV receiver to the ARC-compatible HDMI socket on the TV, but it wasn’t great as the TV downmixed everything to PCM stereo.

Took me less than 10 minutes to set everything up properly (console, Blu-ray and cable box wired to AV receiver with HDMI, then HDMI connection to TV). I also dug out the auto setup mic and re-calibrated the receiver to the new location. He was pretty amazed at how good it sounded.

As a final favour, I ran the THX video calibration program (hard to come by now but it’s on the original blu-ray release of Terminator 2) and correctly set up his rather nice OLED TV.

Curry was on him that night. :-)

Calibration for Pantone spot-color printing can be useful unless P.Inc demand a King’s ransom in licencing fees.

Nothing is free in capitalism.

Adobe Just Held a Bunch of Colors Hostage
Certain Pantone collections now require users to pay $15 a month to access them—with colors turned black unless you pay up.

[ https://www.wired.com/story/adobe-pantone-color-subscription-fee/ ]

So we have a free press as long as we use only black and shades of grey.

Strange there isn’t an Open Source version of a color system covering RGB, CMYK and spots. Which doesn’t make it pretty insane to allow patent COLORS. Bugger Kali Yuga.

The wonderful world of open source gives us this:

“In the Linux world, there has been a move to replace the X11 graphical user interface system with Wayland.

Unfortunately Wayland was developed with little or no consideration for support of color management. In fact it appears that some of the fundamental assumptions made in Waylands development blindly ignore the realities of the display device dependence of color.

Starting in 2013 I attempted to engage the Wayland developers in discussing the challenges of adding proper support for display color management under Wayland, but in general received (and continue to receive) a very hostile response."