TTF Fonts

Post removed. Nemo offered a valid argument.

I can’t imagine why you would want to do this. It’s like trying to convert a 1960s B&W 4:3 TV show to 4K 7.1 Widescreen and wondering why it looks worse than every other freely available content. (Or not wondering because you’re half-blind and never watch TV)

Trinity is a rip-off of Times – Windows has supplied Times New Roman as standard since 1998, and currently comprises nearly 4,000 glyphs across six scripts.

Homerton is a Helvetica clone – Windows has supplied its clone Arial since 1992. Its 4,000 glyphs cover Latin, Cyrillic, Greek, Arabic and Hebrew.

Corpus is a version of Courier – Windows has had Courier New since 92. Currently containing 3,500 glyphs it is capable of Cyrillic, Greek, Vietnamese, Hebrew and Arabic, as well as all the Latin-script languages.

NewHall is Acorn’s version of New Century Schoolbook – Windows has supplied Century Schoolbook since 1994, covering Latin, Cyrillic and Greek.

The Sassoon font is a copyrighted commercial product. I am certain you do not have permission to reproduce a version by that name outside the RISC OS licensed use: Website

These fonts contain a glyph repertoire that barely covers Latin, none of the glyphs are hinted, they all suffer problems that even FontForge is happy to identify, and even if auto-hinted and fixed-up, they’re gimped, cheap knock-offs of vastly superior easily-available fonts produced by actual font foundries that care about utility and quality.

There’s quite enough dreadful fonts in the world. Please don’t add the piss-poor RISC OS ones to the pile.

Valid argument. That depends.
I would not say ‘no use’, but ‘very little use’.

I use Ovation Pro on Windows and sometimes it would be cool to be able to use the standard RISC OS fonts. Other case: PDF generation from a PC, were these fonts will not need to be embeddeded (and so, a smaller file on both Windows and RISC OS). Today I use Open Source fonts to achieve this, but some do not like to install them on limited RISC OS machines.

James, if you still have the archive, I’m a client :)

I use Ovation Pro on Windows and sometimes it would be cool to be able to use the standard RISC OS fonts.

Trinity maps to Times New Roman.
Homerton to Arial.
Corpus to Courier New.
Sassoon to Comic Sans (mostly).

There wasn’t much necessary in converting the issues of Frobnicate written on Acorn to print via Windows. No need to set up RISC OS fonts, just let OP handle the translations. Deals with fi/fl ligatures too. ;-)

Yep, but the rendering (in PDF file) has always some differences.
I guess there will be less problems with TTF versions of the Acorn fonts.
Needs to be tested anyway.

I’m pretty sure any typographer would be pretty aghast by some of the Arial=Helvetica comments in this thread! Even I find them to be quite different, and only have a basic knowledge (ie. my eyes). I’m not saying they don’t make valid substitutions, but they are far from identical. Homerton and Helvetica are much closer, although there are still differences, but probably not enough that anyone would notice.

But then, I just love the RISC OS font manager – always have. It just looks so much better than other alternatives to my eyes, especially in Big Mode (180dpi) – the text looks gorgeous. Now, I’m not saying you can’t get just as nice results with FreeType and so on, but the RISC OS font manager punches so far above its weight, it’s not even funny. IMHO. And I’m biased!

I’m pretty sure any typographer would be pretty aghast by some of the Arial=Helvetica comments in this thread!

In terms of taking something from RISC OS and putting it on a printed page on an alien system, Arial and Homerton (not Helvetica!) are practically the same. Very similar metrics, weight and visual size. A blind idiot translation (this font = that font) will usually suffice. For body text you generally don’t need to fart around with tweaking the point size to get the text to fit again, mess with the leading our baseline or any of that complicated tedious stuff. It generally just works.

However, to anybody with a trained eye, or even an untrained eye if you know what you’re looking for (Arial Bold isn’t as heavy, the font is curvier giving us a ‘c’ that could pass as a yawning PacMan, and it’s messy and all over the place when it comes to the fancy twiddles – there’s no descender on the ‘G’ and what the hell is with that lowercase ‘t’?).

I didn’t say they were identical. I said that Homerton mapped to Arial, with the implication that it was an acceptable substitute where translating the RISC OS font set would not be.

Question for the original poster (no need to delete your post, it’s okay to accept a valid argument!) – what character set mapping was your font? RISC OS isn’t strictly ISO-8859/1 and neither is Windows. It’s that undefined bit in the middle where things like “sexed quotes” lie.

And, for everybody… http://www.ironicsans.com/helvarialquiz/index.php ;-)
(might need scripting, I did it on my phone, got 19/20, Toyota threw me)

It just looks so much better than other alternatives to my eyes

I’m not sure I can agree any more. I used to think that RISC OS was far better than horrible Windows, but I must concede that my Windows is XPSP3. I’m sure a lot has changed in the intervening years (even if Microsoft couldn’t code their way out of an open box); however I’m writing this on my phone. A Samsung S9 with some insanely high screen resolution. The text… Is good, pin sharp. Zoom in and it stays that way. I’m not sure I entirely like this font, but that’s not a fault of the rendering. If I switch to the home screen, the backdrop is of a miko (drawing) and the picture moves slightly as I tilt the phone. It’s all perfectly sharp. No odd colours or aliasing effects, nothing. It’s just a damn good rendering.

For body text you generally don’t need to fart around with tweaking the point size to get the text to fit again

True. The only place I find trouble is labels on diagrams, which sometimes don’t fit nicely when Homerton is converted to Arial. Rather than having a PC/Mac version of Homerton, I convert all the text to path in the diagrams. Sometimes makes the files quite a bit bigger, but that really isn’t an issue these days – it’s not as though I’m dealing with more than a handful of words on diagrams.

No one has claimed that Arial is Helvetica, any more than Homerton is Helvetica. Arial is a metrically identical visual equivalent for Helvetica created to avoid licencing fees, just as Homerton was. Its entire purpose is to substitute for Helvetica without disturbing formatting or visual weight. Saying “I can tell the difference in this glyph’s shape” is entirely missing the point.

The difference between Homerton and Arial is one of level of technology. Homerton is nearer to hot metal than what anyone would call a font these days. Indeed, Helvetica has not only spawned Helvetica Neue, but has now morphed into “Helvetica Now”, adopting things learned from Arial and other Swiss Sans efforts: https://www.monotype.com/fonts/helvetica-now/

the RISC OS font manager punches so far above its weight, it’s not even funny

I ought to let this go but it’s too vastly ill-informed, and that’s not at all funny.

I have a book on font technology to hand: It’s not at all up to date, having been published over a decade ago. It therefore doesn’t cover any of the web font formats, variable fonts, any of the coloured font formats or SVG fonts, and still talks about Multiple Masters (predating my proof of the mathematical flaw at the heart of that format).

The book runs to over 1,000 pages. The Acorn font format would merit no more than a footnote. It hardly qualifies as a font format at all in comparison. It was obsolete 20 years ago.

I don’t expect everyone to be an expert in font technology, but “it looks all right to me” is a disqualifying statement. You’ve no idea what you’re talking about. It’s like someone who has only ever seen a slide-rule saying that computers are over-rated.

Every part of FontManager is inadequate, from the limited glyph repertoire; inability to handle non-idempotent characters; lack of ligatures and combining accents; discretionary features; fixed text direction; to its blurry, partially hinted, gamma-distorted rendering. It should be discarded.

It’s like saying you don’t need WiFi and SSL because you have Econet. I’d use the facepalm emoji but no one on RISC OS would be able to see it.

I’m sorry if I come across as overly fierce, but it’s like the flat earth society in here sometimes.

the RISC OS font manager punches so far above its weight

It wasn’t bad at all. In the late 1980s…

Mind you, I liked hot metal. Where a superior character or a smaller point size character may not be simply a scaled version, but redesigned to suit its size…which I did (in the early 1990s) for our journals, on RISCOS…

Yes, optical size is one of the many hundreds of things of which FontManager is ignorant.

The “Acornthink” is nothing new of course. In the early nineties Acorn tried flogging a “DTP system” comprising an A5000, a laser printer and the awful “Acorn DTP” (which was a port of Timeworks Publisher for the Atari) as a professional typesetting tool for printers. They sent their “Print Sales Manager” to pitch it to me.

“How have you solved the dot-gain problem?” I asked him.

“What’s dot-gain?” he said.

We used Impression, and later Impression Publisher. And printed CRC on Calligraph A3 laser printers, for reduction (62%) on CUP’s process camera. Matched CUP’s typesetting well enough that CUP’s own people couldn’t tell which pages their colleagues had set, and which had been set on the Acorns – we didn’t get up to speed quickly enough to do a whole journal ourselves at first, so we really had to match perfectly.

Fine for text and line diagrams – hopeless for half-tone images :-(

It’s like someone who has only ever seen a slide-rule saying that computers are over-rated.

I used to know a guy (one of those “when I were a lad” types) who probably would have that exact opinion.
And would be completely serious about it.

Last I knew (around 1995) he was complaining about so many “computerised codes” (barcodes) on products because checkout girls were too thick to be able to read a price label. Which led on to a mostly incoherent rant about the state of education and how those new exams (GCSE, not exactly new by then) were nothing like the good old solid A levels (uh, they weren’t, they mostly replaced the O levels).
I dread to think what he’d make of the world today. Everybody so obsessed by this imaginary world in their hands…

Every part of FontManager is inadequate,

To give it some credit, it was born in a time when Windows wasn’t a known thing, when the most popular word processors of the day didn’t even run in a graphics mode (except for a page-on-the-screen preview) and most people with laser printers used something that mapped to the printer’s internal character sets rather than attempting to do it via raster.

To be sure, big impressive and powerful DTP systems did exist but they were eye wateringly expensive, sometimes more than the already expensive computer that they ran on. Ways to do all sorts of fancy things, CMYK and spot colour, accents and glyphs and random foreign type had been solved in various manners.

What was kind of new was to have this sort of thing on the desk of a home user, in a kid’s bedroom, in a domestic setting. Not a big expensive machine that controlled a giant offset printer or a linotype.

from the limited glyph repertoire;

I’m not really sure Acorn (and a lot of their userbase) really saw beyond the shores of their scepter’d isle. Just look at the lack of support for anything other than bog standard English, to the abysmal mess that calls itself the manager of Territories.

Also, to be fair, Unicode wasn’t really a thing back then so assumptions were made regarding mapping fonts into various 8 bit character sets Latin1, Latin2, etc etc. The rest was sort of bolted on later.

However sort of at the same time, over eastwards, text was maybe one of the half dozen incarnations of ISO-2022-JP, or maybe Shift-JIS, or maybe something else. I dimly recall the early MFC classes (on Windows) having some sort of double byte wide text that wasn’t UTF-anything, and maybe this lingers on in the mostly ignored wide character functions in CLib?
Indeed, one could argue that this issue still hasn’t been solved today, what with the different UTF versions in addition to all the old ones (!) and the never ending mojibake that results from incorrect conversions from this to that.

inability to handle non-idempotent characters;

? Google tells me that idempotent means something that can happen multiple times without changing the behaviour differently to it happening once. A classic example would be the guy that spams the button on a lift. Pressing it once registers that somebody wants the lift to stop there. Repeated pressing does not change this, and does not make the lift arrive any faster.

How does this apply to fonts?

lack of ligatures and combining accents;

We have an fi ligature and an fl ligature. ;-)
As for combining accents, yeah, good like getting something like Vietnamese out of FontManager. For those not reading on RISC OS, it’s called tiếng Việt (with both a circumflex and an acute accent on the first ‘e’, and a circumflex and a dot below the second ‘e’). Google tells me an important historical poem is called “Đoạn Trường Tân Thanh”, there’s plenty there to slaughter FontManager, and that’s a Latin based script, not something complicated like ما هي مضيعة حجة جيدة !

fixed text direction

I thought FontManager could write backwards to nominally support Hebrew? But, yes, traditional Japanese goes top to bottom, in columns reading right to left. I think Chinese is similar? And this isn’t even covering the versions of Indian? Arabic? Georgian? that don’t follow a straight line (up or down).

to its blurry, partially hinted, gamma-distorted rendering.

To be fair (yet again) it did give good results when your output device was basically a glorified television. And compared to every version of Windows prior to XP and the introduction of ClearType, it was vastly superior. Those TrueType fonts were really awful to look at. And even then, ClearType had to be “tuned” to give the least objectionable results.

It should be discarded.

FontManager wasn’t bad in it’s early days. It was contemporary with some and ahead of others. The problem is that the rest of the world has advanced, and the internet means that it really isn’t a surprise any more to see stuff like English, ελληνικά, and 汉语 together at the same time. Or wanting to read actual Japanese and not writing it all in kana because that’s all that can be supported. Or the overambitious use of 👻boo! 🦄 Brexit 🐙 tentacles 🥒dildo 💩poop…

FontManager, by contrast, had a sort of half thought out version of Unicode retrofitted, but it’s too dumb to substitute a character that isn’t in a given font with a character that is available elsewhere (so if you want inline עִבְרִית‎ then it’s up to the application to detect that, switch fonts, then switch back) and it’s too stupid to attempt to revert illegal UTF-8 sequences as Latin1 – which alone pretty much guarantees that RISC OS will not sensibly support UTF-8 in the foreseeable future (as switching will automatically break everything written in a western European language other than (drum roll, you won’t guess this…) English.

the awful “Acorn DTP” (which was a port of Timeworks Publisher for the Atari)

Wait, what? Acorn DTP was a port of something?!
Jesus.
It was so bad that I thought it was basically a demo of their new outline font rendering that got botched into being an actual software product flogged to the hapless.

I said it before, will happily say it again. The kid that stole my Acorn DTP floppies from the locked computer room (at boarding school) did me a favour. The school paid out, so I went and bought a real DTP program – Ovation.

[hmm, whatever happened to Tempest?]

Every part of FontManager is inadequate, from the limited glyph repertoire; inability to handle non-idempotent characters; lack of ligatures and combining accents; discretionary features; fixed text direction; to its blurry, partially hinted, gamma-distorted rendering. It should be discarded.

Are you volunteering to write an open source replacement for us then? We can only discard FontManager if someone is prepared to write an improved replacement. Someone who’s an expert in this field.

If there’s no replacement forthcoming, we’re stuck with what we’ve got aren’t we?

the most popular word processors of the day didn’t even run in a graphics mode (except for a page-on-the-screen preview rough simulation)

Fixed that for you.

whatever happened to Tempest?

Wasn’t that the offering that never quite appeared when Impression Publisher and then Ovation Pro took large parts of the RO user market?
Never a large market anyway.

If there’s no replacement forthcoming, we’re stuck with what we’ve got aren’t we?

Ah, but this is a forum. Forums are for whinging…

Nemo offered a valid argument.

What was your original post? I’m guessing you were either proposing to convert RISCOS fonts to TTF, or asking if anyone had already done so.

While creating PDF manuals for Archimedes games, I’ve quite often required an old font that isn’t in TTF. For the most part I can let Acrobat create the font, but in some cases where there’s an image overlaid by a font this isn’t possible and I then have to create the image and find a matching font. On several occasions, having RISCOS fonts in TTF format would have saved me a lot of work.

My http://clive.semmens.org.uk/RISCOS/XP1FontEd.html can convert RISCOS fonts to SFD, which you can then import into FontForge to produce TTF fonts. (It falls over on some more recent ginormous RISCOS fonts but it manages all the usual old ones okay.)

I’m guessing you were either proposing to convert RISCOS fonts to TTF, or asking if anyone had already done so.

More a case of had done so, at which point Nemo pointed out that just because the OS was open source did not mean that the commercial rights on items associated with it (Sassoon font specifically) had been dumped, hence the post deletion I think.
Other portions of Nemo’s response sparked the discussion of the merits of the fonts overall.
That largely amounts to they exist, they work, but you can get better very easily.

On several occasions, having RISCOS fonts in TTF format would have saved me a lot of work.

That’s a rather special case, where you’re trying to reproduce something vintage, and Clive has the solution (although I’m pretty sure there are others)

Clive recalled

We used Impression, and later Impression Publisher. And printed CRC on Calligraph A3 laser printers

The work Richard Pillar did on that device was extraordinary. He went back to South Africa I think. Such a shame. That Toshiba engine was really quite good. 1200×600 with modulated dot size IIRC.

hopeless for half-tone images

It was just-about-acceptable for newsrags.

Rick defended

to be fair, Unicode wasn’t really a thing back then

But it was a thing from 1991. That’s 28 years ago.

I want to be clear about this: Unicode had combining accents in 1991. RISC OS still hasn’t got combining accents in 2019. There’s no defending that.

Indeed, one could argue that this issue still hasn’t been solved today

No, it’s very much a solved problem, just as 3x+4=10 is a solved problem. The fact that some people get it wrong does not mean it is unsolved.

what with the different UTF versions in addition to all the old ones (!) and the never ending mojibake that results from incorrect conversions from this to that.

UTF-8 is ubiquitous except for those APIs that were formulated and frozen in the UCS2 days. Mojibake is rare on computers now. I do see it on devices though – TVs for example – because somebody somewhere doesn’t care.

inability to handle non-idempotent characters

I was using “idempotent” in the “immutable” sense – always the same no matter how, or how many times you use it. This isn’t even the case with simple letters like ‘f’ in a decent OTF font (never mind Arabic), which will produce a few different versions of the glyph depending on the letters around it and the features enabled in the text.

A font like Zapfino has so many different variations that it comprises 1,677 glyphs despite only covering Latin script.

fixed text direction

At the API level – you can’t change direction mid-string, as is required for quoting other scripts. So that has to effected at the app level. FontManager can’t do it.

Andrew joked

Are you volunteering

This did make me giggle, thanks.

What should have been done quite some time ago:

• Port FreeType & HarfBuzz as modules
• Reduce FontManager to the API & caching role
• FreeType & Acorn Font Format supported as glyph provider subcomponents
• HarfBuzz supported as shaper subcomponent
• API clarified and extended
• Composite/virtual fonts added

The work Richard Pillar did on that device was extraordinary. He went back to South Africa I think.

Yes. Their operation was just round the corner from ours, so we became quite good friends. Yes, he went back to SA, and sadly we lost touch. The first version of their Toshiba was 600×600 with modulated dot size, the later version was 900×900, also with modulated dot size. The improved engine came just in time for us – our editorial board decided to increase the journal page size to A4, so we weren’t reducing so much.

RISC OS still hasn’t got combining accents in 2019.

No, the OS hasn’t – but with a bit of judicious fiddling we had them in Impression in 1992. All automated so the copy-editors didn’t have to know what was going on under the hood, but a horrible hack…

Uggh. I can guess exactly how you ‘hacked’ combining accents, and I want you to know I’m not angry, just very, very disappointed. ;-)

Needs must when the devil drives, nemo. The end result looked okay, you might be surprised to learn – they were zero width (as you’d expect) and preceded the letter they belonged to, and then vertical and horizontal kerning was applied before and after them to shove them into position. Straightforward enough for the ones that arrived with the author’s file and were processed into DDF by the input filter, more of a faff for any the copy editors put in later in Impression – had to export text containing any accents as DDF, drop it onto my little Accents app on the icon bar, and drop it back into Impression.

(Worst ones were those over l.c. I – needed a kern before the i as well as each side of the accent. And of course different kerning according to which font.)

vertical and horizontal kerning was applied before and after them to shove them into position

“I felt a great disturbance in the Force, as if millions of font tables suddenly cried out in terror and were suddenly silenced. I fear something terrible has happened.”