Re: Experiments with classical Greek keyboard input
I'm sorry for not contributing more to this thread or to the work needed to be done to solve the issues discussed here... But here are some thoughts in response to Kostas' message: From: Πιστιόλης Κωνσταντίνος pistiolis στο ts τελεία sch τελεία gr This keymap defines a dead key for every combination, and is more or less followed by the windows XP, using up to 16 or more dead keys! You know, there really should be a way to create a keyboard layout on X11 compatible with the Windows XP / typewriter one. Is this currently possible? To do this, either many more generic dead keys are needed, or a way to have a single keypress produce many keysyms, for use in a compose sequence. For reference, here's the Windows XP way to produce polytonic Greek characters: http://support.microsoft.com/default.aspx?scid=kb;el;GR750052 According to the table there, the dead keys used are [ ] - = | \ / ; ' combined with Shift, Alt, and AltGr. In total, 27 different virtual dead keys... Not an easy system to learn, but I think anyone who's learned it, should be able to keep using it under X11. Is it possible to implement this with the current xkb plus simple Compose-file infrastructure? Or is it only possible with complex input method software? 1. most of the dead keys are too often used to be put in third level (except for makron, vrahy). Each symbol is aproximately used in 1 every 3-5 words! Right! By the way, I've typed a bit more polytonic Greek recently, and the layout currently included in XFree86/X.org worked nicely for me (who isn't used to any other polytonic Greek layout). I don't know if the latter odd combination would produce conflicts in an international Compose file, but this idea was used in the past in greek keyboard, in the following combinations: dead_tonos + . : above (middle) dot dead_tonos + : « dead_tonos + : » I don't think there are any conflicts, and these combinations are very nice from a usability point of view: you don't have to memorize obscure AltGr combinations, just to remember that puting an accent on a character that doesn't take one produces a special (less common) character that looks similart. The three combinations listed above were also used in some old MS-DOS keyboard drivers. The present pc/gr file uses altgr for the euro symbol, the middle dot and the «» symbols, along with the Compose combinations and I suggest the same (duality) for all new symbols Another idea is to use the same kind of rules to increase the usability of the polytonic keyboard for writing tenchical texts: To have a double press of a dead_key and the altGr + dead_key to produce the lost symbol so that the user wouldn't have to ... I agree with this. Another proposed use of altGr is for the dead acute. ELLOT, the Hellenic Standard Organization has proposed and defined different symbols for acute and tonos (which is actually the same symbol) which are equivalent in unicode. That was a mistake... My opinion is that having different glyphs for OXIA and TONOS in fonts is a bug. Upright and slanted oxia don't have any meaningful distinction in Greek, they're just graphic variants. Some fonts are designed with a modern look, where oxia looks like a bullet or an equilateral triangle. These fonts can only be used for modern Greek. Other fonts are designed more traditionally, with a slanted oxia. Putting glyphs with upright oxia in these fonts looks, IMHO, ugly, and I think was only motivated by font creators looking at Unicode, seeing characters both with OXIA and with TONOS in their names, and naïvely deciding to differentiate their appearance, without noticing that they are equivalent according to Unicode, and without a justification in representing actual Greek text. By the way, there is a case where font designers have almost universally drawn two canonically equivalent Unicode characters differently, and that's U+00B7 MIDDLE DOT (·) and U+0387 GREEK ANO TELEIA (·). Here they're next to each other: ·· Most fonts have different glyphs for them, because the usual appearance of middle dot looks wrong as an _ano teleia_. So... in this case there is some justification. But the correct way to solve this according to the Unicode model is with higher-level protocols and smart fonts. For example, with modern smart fonts (OpenType etc.), it's possible to have both U+00B7 and U+0387 assume their correct shape and position depending on their surrounding characters. The combination altGr-dead_tonos + vowel is proposed to produce the letter with accent, in case someone needs it. Well... it probably won't hurt much, except in perpetuating the idea that tonos/accent and oxia/accute are different. And also systems which do their own keysym processing (i.e. GTK+) will have to add some more illogical combinations... I have a question. It is mentioned that it's a bug to use dead_horn and dead_ogonek and that combining comma above 0x0313 and combining reversed comma above
Re: Experiments with classical Greek keyboard input
You know, there really should be a way to create a keyboard layout on X11 compatible with the Windows XP / typewriter one. Is this currently possible? To do this, either many more generic dead keys are needed, or a way to have a single keypress produce many keysyms, for use in a compose sequence. For reference, here's the Windows XP way to produce polytonic Greek characters: http://support.microsoft.com/default.aspx?scid=kb;el;GR750052 According to the table there, the dead keys used are [ ] - = | \ / ; ' combined with Shift, Alt, and AltGr. In total, 27 different virtual dead keys... Not an easy system to learn, but I think anyone who's learned it, should be able to keep using it under X11. Is it possible to implement this with the current xkb plus simple Compose-file infrastructure? Or is it only possible with complex input method software? I thought of this too, but I don't see an easy way to do this with xkb. Anyway, the idea of using combinations of dead keys instead of a dead key for every mark combination was used before in macintosh and as long as the single symbol dead keys have the same position with the old keymap... perhaps it is enough for now. It is propably better to implement this legacy keyboard map with some complex input method at a later time, instead of messing up xkb now. ... I don't know if the latter odd combination would produce conflicts in an international Compose file, but this idea was used in the past in greek keyboard, in the following combinations: dead_tonos + . : above (middle) dot dead_tonos + : « dead_tonos + : » I don't think there are any conflicts, and these combinations are very nice from a usability point of view: you don't have to memorize obscure AltGr combinations, just to remember that puting an accent on a character that doesn't take one produces a special (less common) character that looks similart. The three combinations listed above were also used in some old MS-DOS keyboard drivers. yes, it is a very good idea, but in an international compose file it would be a conflict if greek keymap wanted to use: dead_acute + . : above (middle) dot and some other language's keymap uses: dead_acute + . : degree symbol The dead_XXX definitions are accessible for all languages (and this is correct). The correct way to do this would be to have xkb defining a different Compose file for every keymap ... Another idea is to use the same kind of rules to increase the usability of the polytonic keyboard for writing tenchical texts: To have a double press of a dead_key and the altGr + dead_key to produce the lost symbol so that the user wouldn't have to ... I agree with this. But: 1. it could cause the same kind of conflicts as mentioned above 2. in the proposed keymap dead_horn is placed in ' so we want the rule dead_horndead_horn: '\'' But if someone creates a new keymap with dead_horn placed in ] we won't be able to add a new rule. This will work for only one keymap messing up all the (future) others (if we ever need any) Another proposed use of altGr is for the dead acute. ELLOT, the Hellenic Standard Organization has proposed and defined different symbols for acute and tonos (which is actually the same symbol) which are equivalent in unicode. That was a mistake... My opinion is that having different glyphs for OXIA and TONOS in fonts is a bug. Upright and slanted oxia don't have ... are equivalent according to Unicode, and without a justification in representing actual Greek text. ... is some justification. But the correct way to solve this according to the Unicode model is with higher-level protocols and smart fonts. For example, with modern smart fonts (OpenType etc.), it's possible to have both U+00B7 and U+0387 assume their correct shape and position depending on their surrounding characters. I agree The combination altGr-dead_tonos + vowel is proposed to produce the letter with accent, in case someone needs it. Well... it probably won't hurt much, except in perpetuating the idea that tonos/accent and oxia/accute are different. And also systems which do their own keysym processing (i.e. GTK+) will have to add some more illogical combinations... I could hurt because many people will prefer to use it, in order to avoid this bug of the fonts. (and this will cause a lot of trouble when mixed up with monotonic greek of a linux with hellenic locale) This is why I propose altGr-dead_acute, so that the combination will be hard, forcing people not to use it. Unfortunately this is necessary, because a lot of polytonic greek texts are encoded like that. If you want to search text with google you will have to use this accent. Look at google search results. Searching for: ἀνθρώπου (with tonos) yields 584 results and ἀνθρώπου (with polytonic set's acute) yields 21.400 results! (I think that this happens because most texts are converted from older 8bit encodings) This is a google bug (?) too,