Re: [AD] From AL: Dead keys in allegro/mandrake

[Peter Wang <tjaden@xxxxxxxxxx>]

Christer Sandberg wrote:

> On Saturday 05 June 2004 10.25, Elias Pschernig wrote:
>  
>
> > Now, we leave X11, and those Allegro PC HW scancodes are passed on to
> > the DOS/windows keyboard handling, which does another mapping from
> > scancodes to Allegros KEY_* codes, and those are used to get the
> >    
> I think it is correct to name at most one of the two codings `scancode'. 
> Unfortunately docs and examples names the latter one scancodes and also in 
> some places in the source (variable names comments etc. in e.g. keyboard.c). 
> The former coding is also called scancode in the source (in e.g. xwin.c).
> I guess that `pc-hw' is the one that should be the real scancode, if I 
> remember right it was in that context it was used. People that knows about 
> that might be confused when reading Allegro docs. If changing the docs, 
> people that has learn by using Allegro might be confused. People that reads 
> the sources will get confused anyway... I know that for sure.
> Hmm, don't know what would be the best to do about this.
>  

I think 'scancode' should be reserved for the codes produced by the 
keyboard hardware, and emulations thereof (i.e. pckeys.c).  Allegro's 
KEY_* constants can be named 'keycodes'.

Internally, platform-specific codes can be called "X keycodes", etc.  
This is just to reduce confusion for people reading the source code though.

> Another confusing thing (that has nothing to do with the X-problem) is
> the KEY_* names. Obviously KEY_A don't mean "the key with an A on" but instead 
> "the key in the upper leftmost corner among the letter keys". So names that 
> associates with this would be less confusing, e.g. KEY_LETTER_ROW1_1, 
> KEY_LETTER_ROW1_2 etc. But it is not simple to define the term "letter keys", 
> the keys that have letters varies, depending on language layout, and the 
> names of other keys are not very good either, e.g., KEY_EQUALS is not good to 
> me because the '=' is on shift-0. It would on the other hand
> also be a bit stupid to name things like KEY_LETTER_ROW1_1 (matrix index...),  
> so maybe just use a sequential numbering (e.g. keeping the one that exists) 
> and in the docs present a keyboard layout that shows the numbering 
> (developers that want to use acronyms can do their own, national specific or 
> not)?
>  

The current system seems to be the best compromise.  I assume most 
people know roughly what a US ASCII keyboard looks like, so KEY_EQUALS 
tells you *roughly* where the key would be, better than KEY_62.

If you want sequential numbering, you can [almost] use it already -- 
just write bare integers instead of KEY_*.  Strictly speaking, you'd 
need a guarantee that the KEY_* values won't change in future versions.


PS. I pretty much agree with scrapping the X keyboard code.  Under X:
- just don't bother to try and match the KEY_* codes to physical keys;
- don't use pckeys.c layer;
- readkey() should just return whatever characters that 
XLookupString(xke, ...) where xke is the XKeyEvent given to us by the X 
server.  This will only work for Latin-1 characters.


cheers,
Peter