[Trac #418] Keyboard layout collector ticket.

[Zarro Boogs per Child]

#418: Keyboard layout collector ticket.
-----------------------+----------------------------------------------------
 Reporter:  jg         |        Owner:  warp    
     Type:  defect     |       Status:  new     
 Priority:  blocker    |    Milestone:  BTest-1b
Component:  keyboards  |   Resolution:          
 Keywords:             |  
-----------------------+----------------------------------------------------
Comment (by khassounah):

 Replying to [comment:7 dilinger]:
 Although scripting sounds like a good idea, but kayboard configuration
 have key to characters mapping information *and* keybaord behavior
 information. xkb configuration files captures both pieces of information
 well, the excel sheet doesn't, because that wasn't its purpose as I
 describe below.

 I am pro-automation, but I think this is one of the cases where we will be
 re-inventing the wheel if we try to fix our excel sheet to capture all the
 info and then generate xkb files. Lets just describe then in xkb. It will
 actually be easy to extract the subset of information needed and generate
 the excel sheet with the layout and not the other way around.

 > However, there are enough differences from what we're generating
 compared to what other people have uploaded

 That is because different languages work in different ways. And the
 generic format in the excel sheet doesn't capture that (it was designed to
 build the keyboard print and doesn't include behaviour information). For
 example, the nigerian keyboard will be equivalent to english for the most
 part, but with the accents applied using alt-gr after the base letters are
 typed. For Arabic and Thai (Libya and Thailand) the whole keyboard needs
 to switch to the new language because all the keys are different. So
 totally different symbol files will be used when you're typing english
 versus thai or arabic (and the language key should be used as a short cut
 to switch languages).

 > 1) The keyboard images and xls for (for example) Libya shows both
 english characters and arabic on the same key; ie, 't', Arabic_feh, and
 Arabic_shadda all share a key.  The currently uploaded file (as well as
 upstream's 'ara' mapping) doesn't include the 't' portion; only the
 Arabic_feh and Arabic_shadda portion.  Do we want to include the 't' in
 our mapping?

 The purpose of that spreadsheet is showing the layout of the characters to
 be printed on the keyboard, it doesn't map easily to xkb files. Please
 read above.

 > 2) The xls repeats key mappings; for example, we see the following in
 the output from  our script:
 > {{{
 >     key <AD03> { [ e,   E,      EuroSign,       EuroSign ] };
 >     key <AB09> { [ period,      greater,        Arabic_zain,    period ]
 };
 >     key <AC12> { [ backslash,   bar,    backslash,      bar ] };
 > }}}
 > Does listing EuroSign twice make a difference?  Do we need to repeat any
 or all of these values?

 when the keybaord is english mode, the us mapping will be used, but when
 in arabic mode, pressing the AB09 key generates the Arabic_Zain character
 and pressing it with shift generates a period.

 > 3) AC12 above is the 12th key in the middle row (the asdf row).  The
 current mapping for Libya lists it as BKSL instead.  Which one is correct?
 At least on my thinkpad keyboard, BKSL is the 13th key on the top row
 (qwerty).  We've worked from the assumption that rows have 12 keys, and
 the rest are special keys.

 It is the same key that holds the backslash/bar characters on the us
 mapping. If it is not AC12 there, then it is a bug in the Libyan keyboard
 and we should fix it. But it is in the same location as on the US keyboard
 on the same spreadsheet.

 > 4) Unicode stuff is.. interesting.  What Zephaniah did for our script
 was to have things like 0x01000308.  What I see in the ng mapping is
 U0300.  What I did for pt was 0x00ba.   Does it matter?
 > 5) Zephaniah mentioned something about us missing dead keys, or having
 screwed that up somehow.  Zephaniah, could you please expand upon that
 when you get the chance?

 Those are Zephaniah's...

-- 
Ticket URL: <http://dev.laptop.org/ticket/418#comment:8>
One Laptop Per Child <http://laptop.org/>