View mode: basic / threaded / horizontal-split · Log in · Help
July 07, 2004
Re: obtaining keyboard scan codes
In article <ccg8g9$2sqb$1@digitaldaemon.com>, Arcane Jill says...
>
>In article <ccff0l$1os5$1@digitaldaemon.com>, Andrew Edwards says...
>>
>>The C function getch() provides a means to access most control 
>>characters from the keyboard,
>
>A brief interjection about technical jargon here. If a keystroke generates an
>ASCII control code (a single character code in the range 0x00 to 1x1F
>inclusive), then it CAN be detected by getch().
>
>For historical reasons, certain keystrokes do generate ASCII control codes
>(characters in the range 0x00 to 0x1F). However, not all keys can be guaranteed
>to do this. Many, such as the cursor keys, for example, send "control sequences"
>to a console (and that's assuming a command-line app which HAS a console).
>
>The problem is not that you can't detect control codes, it's that many keys
>don't generation them. More often than not, pressing a key will generate some
>VT102 and ECMA-48/ISO 6429/ANSI X3.64 terminal control /sequence/. And of
>course, the whole process is ENTIRELY platform dependent.
>
>In short, getch() cannot be guaranteed to return a single character from each
>key.
>

Which is exactly why I requested this information! I'd like to implement a
UNICODE version for D.

>
>
>but how do I go about obtaining the ones 
>>it does not cover?
>>
>>Specifically these:
>>
>>enum {
>>   KEY_HOME    =  ##,
>>   KEY_UP      =  ##,
>>   KEY_PGUP    =  ##,
>>   KEY_LEFT    =  ##,
>>   KEY_CENTER  =  ##,
>>   KEY_RIGHT   =  ##,
>>   KEY_END     =  ##,
>>   KEY_DOWN    =  ##,
>>   KEY_PGDN    =  ##,
>>   KEY_INSERT  =  ##,
>>   KEY_DELETE  =  ##,
>>   KEY_F1      =  ##,
>>   KEY_F2      =  ##,
>>   KEY_F3      =  ##,
>>   KEY_F4      =  ##,
>>   KEY_F5      =  ##,
>>   KEY_F7      =  ##,
>>   KEY_F8      =  ##,
>>   KEY_F9      =  ##,
>>   ...
>>}
>>
>>Thanks,
>>Andrew
>
>First off, this is not a D question. You have exactly the same problem in any
>language. The answer, however, is that there is NO WAY to achieve this in a
>platform-independent way.
>
>For a start, not all keyboards actually HAVE the above keys.
>
>For another thing, not all keystrokes send single bytes. Some send multiple
>bytes.
>
>For a third thing, a "scan code" (the term used in your subject title) is not
>EVEN the same thing as what I assume you intended by the phrase "control
>character" in your opening paragraph. A scan code is a hardware number (which
>will ALSO differ from keyboard to keyboard, and from operating system to
>operating system). The bytes emitted up through the event queue (what you called
>"control characters") are a software concept. For example, SHIFT and A will have
>two separate scan codes, but the single character 'A' will be percieved by
>getch(). In theory, of course, any key COULD be programmed to act like a shift
>key.
>
>And for a fourth thing, not all computers even have keyboards!
>
>Sorry if this reply is a bit of a downer, but you have at least identified an
>important area where D-in-the-future might need to go.
>
>It occurs to me that a std module to address exactly this area might be a very
>good thing. Such a module could define an enum (at LEAST sixteen bits wide, but
>let's make it 32 to be safe) containing an enumerated value for every possible
>key on every known keyboard in the world (including Chinese and Russian
>keyboards, and maybe even including those mad extra keys you get these days like
>"LAUNCH WEB BROWSER"?). Such a module could provide a getch()-like function.
>
>But it's time to drop the old way of thinking that each valid key combination
>must generate exactly one ASCII character. D doesn't even /use/ ASCII, except as
>a subset of Unicode, and - in Unicode - the meaning of characters 0x00 to 0x1F
>is undefined (apart for whitespace and linebreak properties). Those ancient
>"control codes" to which you refer hark back to the days of punched paper tape,
>when if you hit control-H on your keyboard, a physical bell would ring on the
>teletype. We're way past that paradigm now, so expecting it to still work just
>like it used to is asking too much. There are just too many keys on a keyboard
>for that to be feasable, these days (and that's ignoring things like Input
>Method Editors, and so on).
>
>It's not a D problem, it's a problem inherent in all computer languages. But
>maybe, if someone has time (and I don't), it could be D that provides the
>comprehensive solution. We are already providing that solution with OUTPUT
>(Unicode). Maybe we could provide a similar solution for INPUT.
>
>Just my thoughts. My apologies to Andrew for being completely unhelpful.
>
>Arcane Jill
>
>
July 07, 2004
Re: obtaining keyboard scan codes
In article <cchajm$1c8n$1@digitaldaemon.com>, Walter says...
>
>
>"Arcane Jill" <Arcane_member@pathlink.com> wrote in message
>news:ccg8g9$2sqb$1@digitaldaemon.com...
>> It's not a D problem, it's a problem inherent in all computer languages.
>But
>> maybe, if someone has time (and I don't), it could be D that provides the
>> comprehensive solution. We are already providing that solution with OUTPUT
>> (Unicode). Maybe we could provide a similar solution for INPUT.
>
>I've been through this exact problem with multiple machines. For DOS,
>Windows, and Linux, the solutions are in the microEmacs source downloadable
>from the front page www.digitalmars.com.
>
>

Thanks! I'll check it out as soon as I get a chance.
July 07, 2004
Re: obtaining keyboard scan codes
In article <cche3t$1h6i$1@digitaldaemon.com>, Andrew says...

>>In short, getch() cannot be guaranteed to return a single character from each
>>key.
>
>Which is exactly why I requested this information! I'd like to implement a
>UNICODE version for D.

Oh Cool! Be the one.

Just remember though that Unicode only defines characters, not keys. So,
returning (say) a Russian letter from a Russian keyboard would fit nicely within
your paradigm, but if the user presses CURSOR-LEFT or F10 or
NUMERIC-KEYPAD-PLUS, there will actually be /no/ Unicode character defined to
represent these keys.

You can't use unassigned codepoints, because what is unassigned in Unicode today
won't necessarily be unassigned in Unicode tomorrow.

You can't use private use characters because that might conflict with someone's
private mapping.

So I'd suggest inventing your own standard. And I'd suggest you make use of
numerical values greater than 0x110000 for keys which don't map to a character
(to avoid conflicting with Unicode. Just remember to document that the return
value is a dchar.

Good luck, and I'll be really looking forward to what you come up with.

Arcane Jill
July 07, 2004
Re: obtaining keyboard scan codes
Arcane Jill wrote:

> In article <cche3t$1h6i$1@digitaldaemon.com>, Andrew says...
> 
> 
>>>In short, getch() cannot be guaranteed to return a single character from each
>>>key.
>>
>>Which is exactly why I requested this information! I'd like to implement a
>>UNICODE version for D.
> 
> 
> Oh Cool! Be the one.

I can only promise that I will give it my best shot... Unfortunately I'm
a very slow learner. Me being here should in no way suggest that I'm an
educated or experienced programmer. As a matter of fact my experience is 
seriously lacking, and my education stems mostly lurking around this
newsgroup for a little over. I'm simply here because I am steadfast in
my desire to learn how to program.

> Just remember though that Unicode only defines characters, not keys. So,
> returning (say) a Russian letter from a Russian keyboard would fit nicely within
> your paradigm, but if the user presses CURSOR-LEFT or F10 or
> NUMERIC-KEYPAD-PLUS, there will actually be /no/ Unicode character defined to
> represent these keys.

Got it! I download the Unicode 4.0 manual and am in the process of
reading. I'll also be relying heavily on Walter's microEmacs for
inspiration.

> You can't use unassigned codepoints, because what is unassigned in Unicode today
> won't necessarily be unassigned in Unicode tomorrow.
> 
> You can't use private use characters because that might conflict with someone's
> private mapping.
> 
> So I'd suggest inventing your own standard. And I'd suggest you make use of
> numerical values greater than 0x110000 for keys which don't map to a character
> (to avoid conflicting with Unicode. Just remember to document that the return
> value is a dchar.
> 
> Good luck, and I'll be really looking forward to what you come up with.

Thanks for the encouragement...

> Arcane Jill
> 
>
Next ›   Last »
1 2
Top | Discussion index | About this forum | D home