Author Topic: Multikey autorepetition  (Read 1223 times)

0 Members and 1 Guest are viewing this topic.

Offline aeforeve

  • Thread Starter
  • Posts: 1
Multikey autorepetition
« on: Fri, 06 August 2010, 04:43:17 »
sory for my english

Multikey autorepetition in linux and x-win

Has decided to write for this purpose - that most to understand - whether is it in linux a defect or and has been conceived

The basic question in that is that if to press 2 keys - why one repeats only?

Will tell I in a spreadsheet I press downwards and to the right - and in general that I expect that the cursor will run on a diagonal - but it go only in one direction

Why? It is a defect or and should be?

I for myself have found in a kernel and in x-win places that for it answer - they strangely enough are almost identical that in a kernel that in x-win

And the theory is that - if not to concern ps/2 keyboards - that an essence here in what usb the keyboard sends in a computer of purely event of pressing and unpressing keys - and it is 8 byte package - 1 byte a condition of keys of modifiers - 2 byte a reserve - 3-8 bytes codes of the pressed keys
For this reason usb the keyboard theoretically to serve only 6 simultaneous pressing (without modifiers)

here - a car usb the key at all is not engaged in repetition - in system that in a kernel that in x-win - for autorepetition the usual timer answers - which joins by key pressing - and through the necessary intervals adds pressing
But it is made so - that autorepeat only one key

Here and at me the question is a defect or and should be?

If interestingly I have made patch for x-win for real multikey autorepetition

Offline Findecanor

  • Posts: 4587
  • Location: Koriko
Multikey autorepetition
« Reply #1 on: Fri, 06 August 2010, 06:35:35 »
Good stuff. Thanks!

I think that you should make this behaviour enabled in the config file. Then, you may have a better chance of getting included into the main branch of Xorg.
Man must shape his tools lest they shape him
-- Arthur Miller

Offline dfj

  • Posts: 171
  • Location: Canada
  • Visit our irc: #geekhack on Freenode!
Multikey autorepetition
« Reply #2 on: Sat, 07 August 2010, 06:30:52 »
Quote from: aeforeve;209770
sory for my english

Here and at me the question is a defect or and should be?

If interestingly I have made patch for x-win for real multikey autorepetition

No problem on the English: your intent is clear.

The original PCs, and the typewriters before them (those which had the feature at all) had autorepeat set to disable as soon as any other key was pressed. Even when that key was released, the auto generally did not turn on.

USB keyboards have the repeat handled on the host, (PC, linux,  mac, etc..) but the host is expected to follow the traditional behaviour by default.

I am torn as to whether the default could have been much different. Some typists don't release the keys quickly enough when typing certain words. If the autorepeat timer was not reset with each new keypress, then a few characters might repeat some distance into a word unexpectedly.
  I set my repeat timeout pretty fast, and the rate relatively fast. I can repeatedly hit a key at over 20 Hz, so the autorepeat needs to be faster than that to be useful.
  However, with that quick timeout, I get repeats I don't want sometimes - especially when very drunk or tired.
  Also, there was a limited amount of RAM on the early microcontrollers for keyboards. Having a timeout on more than one key would consume about 2 bytes for each each key worth of simultaneous auto ability.
  I even own a keyboard without a microcontroller - there is no firmware, the scanning (and this one actually scans) is driven by the external terminal.
  Anyway, that's my theory about the history. I don't got hard facts for you, though. :(

  As for whether the option should exist - well, it is linux: if someone wants it, then the option should exist. Where in the giant config hierarchy of these systems and how it should be configured I do not know. Whether it should even merged into the normal install, rather than maintained as a compile-time option or even completely external patch is something else you will want to talk to the x folks about.

enjoy, and good luck,
Fave Switch manus: IBM, Topre, Matias, ...