UTF-8?

I don’t currently emit combining accents because so few systems cope with them well: Compare the precomposed Ō with the combined ̄O for example.

Indeed. On RISCOS I was able to write that “tidying up” app that took an Impression Publisher file, and adjusted (using kerning, horizontal and vertical) all the combinations (amongst other things). The idea of writing an app like that on any other system boggles my mind. (Yes, it adjusted the combinations, rather than substituting precompositions – we could only have 223 simultaneously accessible characters in a font at that time.)

Hmm… Do I even have an area of expertise?

Oh yes. Wander off and have a think about it.

Is it actually Acorn Font Format that has a concept of horizontal text, or is it the apps?

Neither the Acorn font format nor the FontManager API have any concept of vertical text. It only “recently” gained a Right-to-Left flag (in geological Acorn time).

what do you mean by a stroked font vs a filled font?

The “outline” fonts most people are familiar with are filled outlines. The precise pixels that count as “inside” the outline and the manipulations of the outline before rasterisation vary, but the key point is that the vector path delineates inside from outside.

A stroke font on the other hand is stroked, not filled. So it can only produce glyphs with a single stem thickness. The original Adobe Courier was a stroke font, but soon became an outline font. Stroke fonts are still used (especially in TTF) in resource-limited platforms for CJK fonts – such as phones. Or at least that was the case, but since phones became supercomputers that cunning byte-saving wheeze is seen less often (and was always proprietary – see Agfa Monotype for a notable implementation).

Well, my phone does ruby. It just doesn’t do it correctly (unless your markup is wrong?).

Interesting! It’s supposed to look something like this:

a stroked font IS filled

No, it isn’t.

Acorn Font Format fills the outline but does NOT stroke it

The Acorn font format fills closed contours and strokes open contours, subject to the skeleton pixel size limit.

PostScript strokes the outline with a “thin” (theoretically a single pixel wide) line

No it does NOT, it absolutely does not do this.

Other systems also stroke the outline, I guess (but don’t know) also with a “thin” line

No, they don’t. Name a system that does.

This is why Acorn fonts print badly when naïvely converted to other formats – they’re emboldened, typically by about one pixel

This is wrong. When printing PostScript Acorn fonts are typically converted to Type3 which is subject to the PostScript pixel touching rule (paint all that are touched). Crucially, such converted fonts are totally unhinted, which produces stem distortion, stem inconsistency, pimpling and baseline/x-height misalignment.

However, another problem to throw into the mix is that Acorn users typically do not have calibrated monitors, consequently are subject to the usual 2.2 gamma which neither the FontManager nor ColourTrans allow for, resulting in the text on screen failing to match the actual geometry printed on a high resolution printer. Whether the result looks too bold or too light depends on whether it’s black-on-white or a WOB.

The reason PostScript (and others) stroke the outline is that otherwise thin strokes (sorry about this dual use of “stroke” – not my fault!) and serifs in many fonts can drop out completely at small sizes

More nonsense. PostScript does NOT stroke fonts (unless you use charpath, but then the result of stroking that is unlikely to match the hinted filled glyph created by the font renderer, and you get gaps).

Dropout is not a problem for Type3 because of the PostScript pixel touching rule that I’ve only explicitly explained about three times so far. Every pixel that any part of the outline goes through, gets drawn. There is no stroking.

For Type1 however it’s a different story – it uses a different pixel touching rule and dropouts and isolated pixels are a real problem, which is why Type1 has sophisticated hinting and Type1 rasterisers have complex and unpublished heuristics to prevent these artefacts, pimpling, etc.

This is the one area where AFF is actually better

Good God. You really think the Acorn font format is the only format with dropout control and hinting? Are you insane or just trolling me now? Read up on Adobe Type1. Read up on the TrueType grid-fitting and hinting system.

not really of so much importance in these days of very high resolution printers

Agfa, Adobe, Linotype, FontLab, and several thousand professional fontographers disagree with you.

Listen Clive, you’re an enthusiastic RO user and slightly knowledgeable about some aspects of Acorn font creation. However, you know nothing of any other system. Worse than that, you think and claim that you do!

I’ve lost count of the number of totally incorrect claims you’ve made. Claims you have not only not retracted but have continued to repeat as though they are fact. Will you please stop. You are contributing nothing but FUD.

There are four stages of competence, and you are gloriously and steadfastly at stage one. Please stop now.

There are four stages of competence

Some say five. Interesting idea overall, but generalisations do tend lose a little.
DK is often quoted for emphasis.

You really think the Acorn font format is the only format with dropout control and hinting?

No, of course I don’t. But it works in a different way. Okay, I misunderstood the nature of that different way, interpreting the observed effects of translating fonts incorrectly.

However, you know nothing of any other system.

I’ve admitted all along that I don’t know much about them, but I know enough to be able to write software to translate AFF font files into Spline Font Database files, which FontForge can then convert into TrueType or Postscript fonts; and I know how to use FontForge to edit those. Those fonts then work on a PC and on a Mac. That’s as far as I go.

Claims you have not only not retracted but have continued to repeat

Sorry about that, not intentional – I didn’t see your earlier posts when I replied to your later ones. I’ve only been dipping in and out between distractions at home.

Amusingly the four stages don’t always exist in all cases – stage one doesn’t apply to riding a bike for example. No one believes that they can ride a bike despite demonstrably failing to do so.

Unless they have serious issues with the definition of the word “riding”.

Riding a bike is a good example, and an odd one for someone like me. I couldn’t ride a bike until I was about 11. I couldn’t do it until I’d worked out exactly HOW balancing a bike works, whereas most folks acquire the skill without ever understanding exactly how it works – indeed many people refuse to believe it works the way it does. Interestingly, most folks who can ride a monocycle (I can’t, although I understand the mechanics of the process) DO understand how it works.

Another favourite of mine is the mechanics of a cat’s twisting in the air to land on its feet. I strongly suspect that cats acquire that skill without understanding the mechanics…

I’ve admitted all along that I don’t know much about them

Then please don’t write authoritative sentences such as those I quoted and refuted.

It’s fine to speculate in the absence of actual knowledge, but it must be described accordingly. Making assertions that are actually guesswork based on limited data is bad enough down the pub, but put on permanent record, indexed, searched and crosslinked, such statements can confuse or grossly mislead an unknown number of people for an extended period of time, totally unnecessarily.

You know you’ve just made it up, but readers have no choice but to take you at your word.

You know you’ve just made it up

I do now, but it’s been my “wisdom” for so long that I’d forgotten – my interpretations of things worked for me and stuck in my head. I shall go back and edit. That may leave some of your posts rather dangling; you might want to go back and look and also edit after a while when I’ve had a chance to fix things up.

Have you uploaded your updated FontEd anywhere?

If you mean the update of XP1FontEd that can do vertical character “widths”, no, and I probably shan’t. It’s currently very messy, just to try the thing out, whereas the existing offering is pretty clean, and I’m far from sure, after everything I’ve learnt from Nemo, that the update would really be at all useful.

XP1FontEd itself isn’t an updated FontEd, it’s a different app, mainly designed to be used alongside FontEd, providing some extra functions.

@ Steve:

DK is often quoted for emphasis.

Dorling Kindersley?

@ nemo:

No one believes that they can ride a bike despite demonstrably failing to do so.

Young children?
I can’t be the only one who was “look! I’m riding my … crash” ?

BTW, thank you for the stoke/fill explanation.

@ Clive:

Another favourite of mine is the mechanics of a cat’s twisting in the air to land on its feet. I strongly suspect that cats acquire that skill without understanding the mechanics…

Are you owned by any cats? If you are, you will know full well that they are likely visitors from another world with minds immeasurably superior to ours, utterly fed up with being talked to like a baby. Their weakness? Having the belly rubbed. Makes ’em all gooey.

I shall go back and edit.

Please don’t.
If there are things you want to unsay, just wrap <s> … </s> around it to strike through the text. This will make it a lot easier for the rest of us who don’t really want to go back and read your messages to try to figure out what it used to say vs what it says now.

@ Chris: Re Ruby text. Hmmm…. Apple 1, Android 0. (^_^)

Please don’t.

Damn. Too late in some cases. Can I undo my edits?

Are you owned by any cats?

No, not since I was a child when the whole family was. When in India, I’m part of a group of humans occasionally temporarily owned by itinerant cats.

they are likely visitors from another world with minds immeasurably superior to ours

Is this speculation, or known fact? 8~)

Re Ruby text. Hmmm…. Apple 1, Android 0. (^_^)

I was bored so I just did a little more testing:

Safari on Mac/iOS – Correct
IE 11 – Correct
Firefox on Windows/Ubuntu – Incorrect; appears as 「漢字かんじ」
Chrome on Windows – Correct
PS4 – Correct
NetSurf on RISC OS – Gotta be joking! :)

It’s bizarre that Android gets it wrong, as I believe that the browser is kinda/sorta like Chrome.

In my defence, I should point out that the effect (on a one-bit-per-pixel printer) of the difference between Acorn’s “paint a pixel if its centre lies within the outline” and PostScript’s “paint a pixel if any part of it lies within the outline” is precisely the same as the effect of stroking the outline with a single pixel wide stroke: it adds an average of half a pixel around the edge, ie an erratic (loss of hinting) emboldening by one pixel – an effect that decreases in significance as the pixel size decreases. This is complicated by the fact that Calligraph’s (Toshiba) printer is not actually one-bit-per-pixel (having variable dot size¹), but the effect of that is similar to antialiasing using a fine half-tone²: visually pretty good but not faithfully copied when the laser print is used as camera ready copy (that infamous dot gain). In practice that wasn’t a problem for text or black-and-white diagrams, graphs etc., but an absolute killer for half-toned greyscales, never mind full colour.

¹ Interesting to look at those variable size dots under a microscope, too. Although they work well as half-toning dots, they’re not hard-edged dots like litho-printed half-tone dots. There’s a fair amount of toner splatter, forming a soft edge – itself a kind of ultra-fine random half-toning, since the actual toner particles are very definitely black.

² Anti-aliasing is, roughly speaking, the logical inverse of half-toning. Think about it. Achieving grey scales on a one-bit-per-pixel grid by half-toning, and then using those grey scales to anti-alias a black-white boundary is crazy: you can get a better edge simply using the grid, which is much finer than the half-tone screen. But the Calli printer is not one-bit-per-pixel.

(Sorry about any bad wording of any of this. I understand what’s going on here very well, but I don’t necessarily remember/know the right words to describe things.)

@ Steve:
DK is often quoted for emphasis.
Dorling Kindersley?

Dunning-Kruger

I’ve just looked and wikipedia seem to have a page that covers the stages and side reference to Dunning-Kruger

BTW. The fifth stage bit is glossed over in the wiki, but many say the 4 stage model doesn’t properly cover the level that the wife refers to as uber-geek

Some say five. Interesting idea overall, but generalisations do tend lose a little.
DK is often quoted for emphasis.

Indeed. I’ve been trying to resist reacting to the gratuitous insult implied, but am failing.

Within my areas of specialization I think I can fairly claim to be at stage 3 or 4 (in many cases, back at stage 3 from stage 4 due to stopping practising, & forgetting terminology). In other areas I know I’m at stage 2. Undoubtedly there are areas where I’m at stage 1, but by the nature of the beast I don’t know which those are (anyone honest has to say the same thing).

The fact that CUP typographers couldn’t tell where in a single issue of a journal the change from CUP’s own typesetting to our in-house Acorn typesetting occurred is, I am sure, adequate confirmation that with a lot of effort from the person who set the system up (me) and relatively very little from the people who operated it (the production editors) we really could match the quality, and that I understood pretty well what I was doing. I knew we couldn’t do half-tones or full colour (although Richard Pillar (at Calligraph) and I did make strenuous efforts, working together, to achieve the former).

I suspect the main reason Nemo said “stage one” was that I said that with very high resolution printers hinting and stem consistency (or words to that effect) was nowadays of less significance. I stand by that. I didn’t say it was of NO significance, but its significance obviously diminishes as printer resolution (and monitor resolution) improves. Obviously we depended on getting AFF scaffolding and skeletons right to achieve what we did in the early 1990s, and I certainly don’t deny that it’s still important. Just less important than it was…

Possibly also that I raised the fact that all the different "f"s in his Zapfino sample could, for all one could tell from the illustration, be in different, very normal, ligatures. Perhaps I didn’t make it clear enough that I certainly wasn’t denying that Zapfino probably nonetheless had plenty of other glyph substitution – just that the sample wasn’t a particularly good illustration of it.

Indeed. I’ve been trying to resist reacting to the gratuitous insult implied, but am failing.

Not from me I assure you.

Indeed not, Steve. The only insults I’ve detected have been from Nemo. I’m not afraid of mentioning it. Nor do I wish to fall out with him – I’m sure I can learn a lot from him, and I’m equally sure I could teach him useful things if he’d let me.

(I’d love to teach him not to be so grumpy and arrogant, but I fear that’s beyond me. At least he recognizes it…)

I only know Sim— I mean, nemo, from these and other RISC OS forums, and both his talents and way with words are abundantly clear. After this encounter, Clive, you may appreciate this joke…

8~)

Dunning-Kruger

Ah. I’ve heard of the Dunning-Kruger effect.

It is rather apt at describing management:

• tend to overestimate their own level of skill – all the time
• fail to recognize genuine skill in others – it is a reason for celebration if you get recognised at all¹, never mind for any particular skill
• fail to recognize the extremity of their inadequacy – an extremity which is about the size of the Pacific
• recognize and acknowledge their own previous lack of skill, if they are exposed to training for that skill. – but training costs money, so that can’t be a good thing

¹ Usual case – something went wrong. Obviously isn’t going to be their fault… suddenly you are “recognised”. In much the same way a sheep feels “recognised” as it is led off to the slaughterhouse.

nemo -

I think once we have the desktop AND the command line supporting UTF-8, it’s worth authors updating their apps to support it too.

If I ask really nicely – would you consider hooking me up with an early copy of the command line UTF-8 handler? [is it a module? kernel source mods?] I would like to verify that my OLED module is quite happy with UTF-8 sequences (I’d likely just test basic kana (if you have implemented it, that is!) – if that works okay, no reason to assume other stuff would fail) and that nothing unexpected happens when providing UTF-8 strings. If there are any specific requirements, let me know.

The way the module works (FYI) is that it accepts a null-terminated string of indeterminate length. Control codes are filtered out (some are recognised, the others are discarded) and the rest is just written character by character with OS_WriteC. All of this takes place while VDU is redirected to a sprite – lets the OS do all the hard work plotting stuff; flip side is that your UTF-8 modifications should be a plug-in-and-go kind of deal. If the OS can do it…..