Compose mode to imput UTF-8

Have you a place where the mods could be downloaded from?

I have now! https://clive.semmens.org.uk/RISCOS/KeyBds.zip – that’s the whole miniapps containing them, but the modules are easily found within. Sadly I’ve still not found a copy of IKHG – I suspect I don’t have one any longer. Someone somewhere probably does…

copy of IKHG – I suspect I don’t have one. Someone somewhere probably does…

Well, about 20 years ago Theo was looking for a modernised one, and did get a reply from both Castle and ROL.
https://comp.sys.acorn.programmer.narkive.com/8v01Bav0/wanted-recent-ikhg

Whether the replies were positive, and he has a 32 bit version, is a question only he can answer.

I would be very interested in a 32-bit version!!

And so would i!

I’ve played with unicode characters and compose systems. AltKeys and MoreKeys look cool, but no source code is a bummer. Check GitHub for similar tools or hit up Stack Overflow for advice.

AltKeys and MoreKeys look cool, but no source code is a bummer. Check GitHub for similar tools or hit up Stack Overflow for advice.

I’ve already wrote a compose system before to be used independently of the X environment for a system
that was going to run also standalone. It’s not complicated. The part I was asking advice for was
about the correct way of capturing the keys and replacing the input for every program. On riscos this can be achieved by claiming the EventV vector and pushing bytes in the keyboard buffer.
Rick’s MoreKeys module does the first part and sets a memory block so the application in the user
space can poll it and push the bytes when appropiate. He shared the source code, and !Chars is a good example about altering the keyboard buffer.

The problem is that I wanted to ignore the system alphabet, but by the way keys and chars are related and how the wimp uses alphabets, applications should
be modified to treat each byte of the keyboard buffer as part of one UTF-8 encoded character, even
when the application supports UTF-8 (like Rick’s zap version) but the system alphabet is not utf8.

I’ve been thinking about it and I can’t help saying to myself: what are you doing?, what the “*” are you doing?

I need UTF-8 so I’m afraid riscos is not for me as things are right now.

the correct way of capturing the keys and replacing the input for every program

The correct way is to write a keyboard handler. That way lies the path of madness…

On riscos this can be achieved by claiming the EventV vector and pushing bytes in the keyboard buffer.

Or within the Wimp environment, using the Wimp_ProcessKey call.
But, note, due to how it was implemented, and how key codes in the range 256-511 have traditionally been used for other things (the Fkeys etc), one must send UTF-8 byte by byte. It’s not possible to hand a 16 bit key code to an app.

but by the way keys and chars are related and how the wimp uses alphabets

That’s exactly it. If you pass ASCII 233, is that an ‘é’ or is it part of a UTF-8 code? Literally the only way to tell is to look at the codes sent at the same(ish) time to see if it’s a valid UTF-8 sequence, which isn’t much help for those apps that don’t understand that byte != character.

Sadly, there is still no built in support for falling back to Latin1 (or whatever) for bogus high bit set characters. Granted, the people that say this isn’t the job of FontManager and/or the purists that say “UTF-8 is UTF-8, end of” may be correct in their assertions, but if there’s no simple sensible way to get old 8 bit characterset apps working under UTF-8, there will not be any compelling reason to ever switch the desktop/system alphabet to UTF-8.

[this, by the way, is why I don’t have much interest in 64 bit anything, there’s still plenty that can be done with the 32 bit software that exists]

when the application supports UTF-8 (like Rick’s zap version)

I think you mean Ovation. ;)

That’s just an epic cheat really. The reason you must hold down Alt when starting the app is in order for it to load its default fonts either with or without specifying the UTF-8 encoding. The only other thing I did was to teach the cursor handler about how UTF-8 sequences are constructed, so when you press the left or right keys, it’ll skip over the correct number of bytes to position the caret at the appropriate place. It’s really simple, and sadly I’ve not figured out the file format enough to know where to poke a flag to say “this file is UTF-8” to get it to load correctly.

I think you mean Ovation. ;)

I also have tried, and like, Ovation. I thought that would be
very, very handy to work with notes mixing typed unicode text
with images. The responsiveness of the software and the mechanisms
for dragging images and edit them again, all with a clean interface.
I really love it. The only fault I found is the lack of undo. This
can be very hard to implement, some times the entire structure of the
program has to be redesign. I thought that you were also responsible
of the UTF-8 support in zap.

The only other thing I did was to teach the cursor handler about how UTF-8 sequences are constructed, so when you press the left or right keys, it’ll skip over the correct number of bytes to position the caret at the appropriate place.