It's the keyboard layout.


A lot of people want a lot of different things.

So lets assume its impossible to make dosbox accept other layouts, contrary to what its manual says.

Would the demand for that warrant a custom run of keymats?

What special game or program is it that needs the level of either US pairs or position each layout claiming to be so works with?

If no, then what is it doing in what i understand to be the main default layout?

Pleasing anyone at the cost of general writing and coding, what audience for a keyboard are you then left to satisfy?
 
A lot of people want a lot of different things.

So lets assume its impossible to make dosbox accept other layouts, contrary to what its manual says.

Would the demand for that warrant a custom run of keymats?

What special game or program is it that needs the level of either US pairs or position each layout claiming to be so works with?

If no, then what is it doing in what i understand to be the main default layout?

Pleasing anyone at the cost of general writing and coding, what audience for a keyboard are you then left to satisfy?
Are you saying we should commit to nothing because we can't please everyone? 
 
I'm not a big fan of _wb_ intl2 getting by with reduced weights for putting labels on keys, and a few other things...  Do you mind if I make a few changes based on that and the PM, _wb_?

It is sad to see fewer people adding their priorities in (thanks btw Urben), but it's also hard when the matrix is changing so much ;) .  You have to go back to make sure your favorites are still in order!
OK, fixed that. I did reduce the count for the labels for éèàçäüö etc (by counting them only as "half" an extra label) because the idea is that those extra labels would be in a smaller/thinner font so they would not distract too much. But in a sense that is cheating, so I went back to fully counting them (which means that people who give negative weight to "clutter" give it a lower score now, e.g. it is no longer your top proposal now, ible :)

I also added the following property:

"German-centric": supports German but not other languages

because we already had a property "German letters labeled (ÄÖÜß)" which I originally gave a negative weight myself, until I started producing proposals that actually have the German letters labeled ;)

My reasoning is as follows: I don't really mind that the German letters (or letters for whatever other language) are labeled (besides for the clutter that it adds, but that is already taken into account elsewhere), but I do mind having a keyboard that looks like it is made for another language/locale than my own or just US-QWERTY. So I don't like a "German-centric" layout (or a "Spanish-centric" layout, or whatever other language), but I do like a layout that either shows no "international" stuff at all, or it has "international" stuff but in a general way, not favoring one specific language like German.

So now I give a positive weight to "German letters labeled" (and also positive weights to the other languages), but a negative weight to "German-centric". In other words: I don't mind äöüß being labeled, as long as éèàçñ¿¡åæø′¸`.¨~ˆˇ are also labeled. I don't know if that makes sense to you guys :)

Here's the link to the big table again, for convenience: https://docs.google.com/spreadsheets/d/14mVNNzLhmhyA4sCIAPSoDv404xtctfLSiEMTtTlEzfw/edit?usp=sharing
 
Thanks ;) .  Yeah I have almost no preference on intl characters, since I wouldn't need them.
 
Saber/wb: Im wondering if there is some group whos wishes you understand to be detrimental to the whole, and what the group you are trying to please in the first place is.

Edit: If the efficiency and familiarity isnt there, it might fail across the board just because each individual user isnt content. The 70% for everyone compromise is very dangerous.

wb: https://en.wikipedia.org/wiki/Tragedy_of_the_commons
 
Last edited by a moderator:
Saber/wb: Im wondering if there is some group whos wishes you understand to be detrimental to the whole, and and what the group you are trying to please in the first place is.

wb: https://en.wikipedia.org/wiki/Tragedy_of_the_commons
The problem with that distortion is "the group" cannot make a decision on it's own, because it doesn't actually even exist, because it's made up of a bunch of individuals with one or more of those said individuals taking the reins(by being the loudest, the biggest, the most influential, and/or the most intelligent) and guiding the rest, "the group" to persuasive action. Without those traits it would be like a tug of war in every direction(like the keyboard discussions for the most part stretched 50+ pages and counting).  ;)
 
First rule when designing something for a big crowd:

KISS

Keep It Stupid Simple

There is no use in trying to make so many different nationalities happy as the trade-offs CK mentioned (efficiency and familiarity) are way more important when dealing with such a big, diverse group.

Unlabeled letters and characters are (imo) a bad idea. Ofcourse the vocal few in this thread that designed the keyboards themselves think it's easy to remember hidden letters and diacritics but the end-user group will think different. Labels are good and eliminate confusion and questions as long as the clutter isn't so bad that it's hard to find what you are looking for.

Which brings me to a 2nd design issue I noticed on many keyboards which is the fact that some of them have symbols and international (other than english) letters spread all over the place which makes finding the right one harder because there is no clear indication of where to look. There might be an idea or system in some of them but in many I can't find it. ;)

Then there's the amount of non english letters on some of the keyboard designs...  First of all, we're not designing a keyboard for a translation machine. Basic international input methods should be good enough, especially when there's room for personal mappings/additions. Most languages don't use more than 4 or 5 "non english" letters anyway. I think it's silly to have 25 of them printed on the keymat when each individual person will probably not use more than 3 regularly and almost every letter can be typed with dead diacritics and/or with a compose key.

Besides that, in my eyes all those language specific letter labels next to the regular letters labels make the keyboards look way too cluttered, intimidating and hard to read.

I know I made 2 dual modifier lay-outs with many (small) labels too, mainly to keep _wb_'s spreadsheet and some dosbox enthousiasts happy but if I'm completely honest, I can't have a commodore 64 lay-out either, I'll deal with it.

The Pyra is it's own machine and thus dosbox or C64 compatibility has a way lower priority. It's not the only software running on the system.

I still think that having everything behind a single modifier is the best and most clear method for every one, even if the double modifier can be achieved with a simple single thumb mash. But if the US standard shifted pairs are very important to have, the dual modifier is necesary as with the limited amount of keys available it can't be solved without it.

KISS!! :-*

8
 
Saber/wb: Im wondering if there is some group whos wishes you understand to be detrimental to the whole, and and what the group you are trying to please in the first place is.

Edit: If the efficiency and familiarity isnt there, it might fail across the board just because each individual user isnt content. The 70% for everyone compromise is very dangerous.

wb: https://en.wikipedia.org/wiki/Tragedy_of_the_commons

The group I'm trying to please in my keyboard layouts are the users of Latin-based written languages. I'm not trying to do anything at all for Greek, Cyrillic, Arabic, Hebrew or Asian writing systems.

For the Latin-based languages (i.e. North and South America and most of Europe), I try to maximize efficiency for as many people as possible, given the constraints of having few keys and little space on the keys for extra labels. I try to treat 'bigger' languages better (though as a side-effect of that, 'smaller' languages can get lucky).

See also:

http://en.wikipedia.org/wiki/List_of_languages_by_total_number_of_speakers

http://en.wikipedia.org/wiki/List_of_languages_by_number_of_native_speakers

Let's make a list of languages. There are different degrees of support, and I'll color-code them:

  • PERFECT: All commonly used letters are available as primary keys  (I'm not talking about the æ in "Encyclopædia Britannica" here)
  • VERY GOOD: All commonly used letters are available as primary or Meta keys (one 'keystroke' if you use the shoulders as modifiers), all diacritics used in the language are available
  • GOOD: All commonly used letters are available as primary or Meta keys or in two keystrokes using dead diacritics
[above: the language can be typed with the letters as labeled on the keyboard; below: remapping or some memorization is needed]

  • OK: One or two commonly used letters are not available, but can be mapped to some other symbol in a somewhat mnemonical way (e.g. the Polish could map their Ł to the labeled symbol ₤; Hungerians could map űő to the labeled üö and use the dead umlaut to get the actual üö, for example)
  • BAD: Three keystrokes (i.e. a Compose sequence) are needed for one or more commonly used letters
  • VERY BAD: It is not possible at all to input this language

This is how my proposal performs the most spoken languages with a Latin-based writing system:

~1200 million people speak English

460 million people speak Spanish

336 million people speak Portuguese

274 million people speak French

over 150 million people speak Swahili

145 million people speak German

75 million people speak Vietnamese (missing: hook above diacritic (Hỏi) and the letters ĂĐƠ, Ư)

73 million people speak Tagalog (Filipino)

71 million people speak Turkish

64 million people speak Italian

40 million people speak Polish (missing: Ł)

28 million people speak Dutch

28 million people speak Romanian

~25 million people speak Scandinavian languages (5M Norwegian, 5.5M Danish, 9.2M Swedish, 5.4M Finnish)

19 million people speak Serbo-Croatian languages (missing: Đ)

~17 million people speak Afrikaans

13 million people speak Hungarian (missing: űő)

11 million people speak Czech

~9 million people speak Catalan (missing: interpunct)

~9 million people speak Quechuan languages

5.4 million people speak Albanian

5.2 million people speak Slovak

3.1 million people speak Lithuanian (needs ū)

2.5 million people speak Slovenian

1.75 million people speak Latvian (needs ū)

1.2 million people speak Estonian

~1.2 million people speak Irish

0.7 million people speak Basque

0.5 million people speak Maltese (needs ħ)

Adding dead strike-through (for Ł, ħ and Đ), dead double acute and dead hook above would help a bit, but in my opinion those are too 'exotic', and those who need them would still prefer to have pre-composed letters instead.
 
What an utterly strange way to denominate dedicated letters as the basic way to type... Completely disregarding both efficiency and habit.
 
The reason why these people voted for it, is because they need it. Very good is not the correct term for something that fails.
 
Even so, you can have something that works better for all of them, and no letters that are impossible to type.
 
Its either compatible with dedicated keys for non-english audiences, or it isnt. I dont see a reason not to.
 
Edit: There is no such thing as scandinavian languages btw. Finnish and the sami languages are different from the rest. Iceland uses a lot of runic letters.
 
Last edited by a moderator:
What an utterly strange way to denominate dedicated letters as the basic way to type... Completely disregarding both efficiency and habit.

The reason why these people voted for it, is because they need it. Very good is not the correct term for something that fails.

Even so, you can have something that works better for all of them, and no letters that are impossible to type.

Its either compatible with dedicated keys for non-english audiences, or it isnt. I dont see a reason not to.
How can you have something that works better for all of them? How would you type, say, Spanish on your keyboard proposal?

If you understand correctly, you would let Spanish people remap [{ to grave/circumflex (like on their keyboards), <> to ñ, keyboard Shift to acute/diaeresis, and perhaps ¡¿ at ]} or maybe as hidden Shift+AltGr combos near ! and ?, which would be Shift+AltGr+1 and Shift+AltGr+6. Then they would map the original symbols <>[]{} to the AltGr-layer, or maybe they would already be available as hidden symbols on AltGr+DFGHJK. In any case they lose their keyboard Shift and they have their language-specific keys not labeled at all, so they are supposed to already know how to blind-type on a normal Spanish keyboard, and then somehow transfer that 10-finger muscle memory to their two thumbs.

You may claim that that works better than having labeled ñ¿¡ and diacritics available, but I respectfully disagree.
 
You cannot go by the number of speakers for a language

1350 million Chinese speakers.... v's "only" 1200 English speakers....

There are clearly some other criteria here!
 
You cannot go by the number of speakers for a language

1350 million Chinese speakers.... v's "only" 1200 English speakers....

There are clearly some other criteria here!
Yes. Purchasing power is probably one of them. Proximity to the physical location of ED's shop is probably also one of them.

But number of speakers is not a completely irrelevant thing either.
 
Here is my most recent and currently best-scoring keyboard layout proposal:

attachicon.gif
wb_intl2.png
Suggestion 1:

Mild key compression

Keys on the Pyra are a precious commodity.  Preserving them, when possible is a good thing.

Move Tab to Fn+Space

Move /? to Fn+I


Move `~ to where Tab was


The 'normal' position for `~ is on the left.  This puts `~ on a key on top line by itself, and can be it's own diacritic under Fn layer. Fn+A is then available for something else. Having tab on Fn+space should not be an issue.  After the Pandora, this feels more natural than having Tab as a dedicated key on a thumb keyboard.  You could add two more diacritics on Fn+A and further widen the language support.

Suggestion 2:

Swap sides on the shoulder modifiers.

Having both Fn+shift modifiers (Keyboard and shoulders) on the same side results in some pretty bad ergonomics when trying to reach the bottom left Fn layer of the keyboard.  This could be greatly alleviated by simply swapping shift and Fn from the left shoulder with Alt and Ctrl on the right shoulder.

Give it some thought?
 
Last edited by a moderator:
Let's make a list of languages. There are different degrees of support, and I'll color-code them:

  • PERFECT: All commonly used letters are available as primary keys  (I'm not talking about the æ in "Encyclopædia Britannica" here)
  • VERY GOOD: All commonly used letters are available as primary or Meta keys (one 'keystroke' if you use the shoulders as modifiers), all diacritics used in the language are available
  • GOOD: All commonly used letters are available as primary or Meta keys or in two keystrokes using dead diacritics
[above: the language can be typed with the letters as labeled on the keyboard; below: remapping or some memorization is needed]
  • OK: One or two commonly used letters are not available, but can be mapped to some other symbol in a somewhat mnemonical way (e.g. the Polish could map their Ł to the labeled symbol ₤; Hungerians could map űő to the labeled üö and use the dead umlaut to get the actual üö, for example)
  • BAD: Three keystrokes (i.e. a Compose sequence) are needed for one or more commonly used letters
  • VERY BAD: It is not possible at all to input this language
So, ~2.3 Billion of the world's ~7.5 Billion people can use it with minimal or no adjustment. This is actually a pretty amazing achievement. Well done. With a few minor refinements, we may have a winner here.

So, for language support, what gets left out?

http://en.wikipedia.org/wiki/List_of_languages_by_total_number_of_speakers

1350 Million Mandarin (Character set incompatible with Western languages)

540 Million Hindi (Character set incompatible with Western languages)

390 Million Arabic (Character set incompatible with Western languages)

376 Million Urdu (Character set incompatible with Western languages)

270 Million Malay (Character set incompatible with Western languages)

260 Million Russian (Character set incompatible with Western languages)

250 Million Bengali (Character set incompatible with Western languages)

150 Million Tamil (Character set incompatible with Western languages)

123 Million Japanese (Character set incompatible with Western languages)

84 Million Javanese (Character set incompatible with Western languages)

80 Million Wu (Character set incompatible with Western languages)

77 Million Korean (Character set incompatible with Western languages)

75 Million Marathi (Character set incompatible with Western languages)

So, from my count, the biggest portion of the world population that is left out of _wb_'s recent proposal isn't going to be able to happen on a single key mat. Per my above post, it may be possible or convenient to add in two more diacritics, which may bump one or two languages up in capabilities on the device.

Overall, nicely done.
 
Suggestion 1:Mild key compression


Keys on the Pyra are a precious commodity.  Preserving them, when possible is a good thing.


Move Tab to Fn+Space


Move /? to Fn+I


Move `~ to where Tab was


The 'normal' position for `~ is on the left.  This puts `~ on a key on top line by itself, and can be it's own diacritic under Fn layer. Fn+A is then available for something else. Having tab on Fn+space should not be an issue.  After the Pandora, this feels more natural than having Tab as a dedicated key on a thumb keyboard.  You could add two more diacritics on Fn+A and further widen the language support.

Suggestion 2:

Swap sides on the shoulder modifiers.


Having both Fn+shift modifiers (Keyboard and shoulders) on the same side results in some pretty bad ergonomics when trying to reach the bottom left Fn layer of the keyboard.  This could be greatly alleviated by simply swapping shift and Fn from the left shoulder with Alt and Ctrl on the right shoulder.


Give it some thought?
I re-thought Suggestion 1.


\| from Fn+ M to Fn+ I key.

/? from Fn+space to Fn+ M key


`~ from Fn+ I to where tab is currently

Tab to Fn + space

It puts \| on P row (where it originates from)

It puts /? on the M row (where it originates from)

It puts `~ on the upper left (where it originates from)  ( Allows Fn+`~ diacritics on it's native key.  Frees up A for 2 more.)

It puts tab on the Fn+space bar (where it worked fine on the Pandora)  Tab on space looks a lot less funky/weird than /? on space.
 
Last edited by a moderator:
I like to have a dedicated Tab, though. On the other hand, what I mostly dislike about the Pandora Tab, is that it is inconvenient to hit it multiple times. With Meta/Fn on a shoulder button, that is less of an issue...
 
I have not been following this discussion closely but looked at some proposals this weekend. The one thing that struck me is that many including _wb_'s tries to be an insane amount of everything at once. Compatible with everything != usable. If each button has so many keycodes that you need a third level modifier just to type in your language, then it is too much to read or print on keymat as well. That means you have to look it up, so why not have more layouts available and switch to a suitable one with limited need for modifiers as the Kingu's functional area? I would rather switch between two sane layouts than hitting modifiers, combobuttons and spend time reading too tiny keymat print.


Cheers
 
Back
Top