plib-users

[Plib-users] Don't Do This!

[Steve B. <sjb...@ai...>]

I'm not a happy person.  :-(

This is not so much a cry for help - as a warning for others in the future:

I've been playing around with a multiplayer game (perhaps going in the
direction of "Super Smash Bros'" on N64 and Game Cube).

I've got to the point of dividing the screen into two, three or four 
chunks and rendering four cute little penguins - each of which can move
independantly with a camera that follows along.  I had all of it working
with the controls for each player assigned from a config file (eg 

  Player 1: MoveNorth = W, MoveSouth = Z, MoveEast = W, MoveWest = S
            Jump = Q, Fire = E
  Player 2: MoveNorth = I, MoveSouth = M, MoveEast = J, MoveWest = K
            Jump = U, Fire = O

...that kind of thing).

So, if you are player 2, you hold down 'I' to move north, 'I' and 'J'
together to move northwest.

It all worked well in testing - but the first time my son and I tried
to actually play together, we noticed that sometimes our keystrokes
would be ignored.  If player 1 moves NorthEast (holding down W and S)
then the 'I' key (player 2's command to move North) wouldn't work.

I puzzled about what might be wrong with my code - but eventually
(after some questions to the LinuxGameDev mailing list and a bit of
web surfing) discovered that this is a hardware limitation of the PC
keyboard!

Aaaaggghhhh!!!!

I hadn't noticed it when testing because with just me pushing one
or two keys, I never hit the combinations that cause problems.

This suprised me a lot (which is why I'm writing this email - so that
others don't fall into the same trap).

The rule is approximately that if you hold down three keys that form
the corners of a rectangle on the keyboards 'scan matrix' (E, C and I
for example) then depending on the age of your keyboard, you'll either
find that the third key you press will be ignored *or* the key at the
fourth corner of that rectangle will appear to be pressed even though
it isn't!

I thought for a while about picking keys that didn't form the corners
of such squares (eg E, C and G work at the same time) - but rapidly
realised that I couldn't come up with anything remotely ergonomic
in that way...not even for two players - let alone three or four!

So now I have to rethink this in terms of two players using joysticks,
one using the keyboard and another using the mouse.

The trouble with that is that *very* few people have two joysticks
and it's hard to make the controls fair when the players have different
controllers.

Grumble, grumble, grumble....

(If you want to try what I've written so far, you can grab it from:

   http://www.sjbaker.org/steve/tuxfighter.tgz

...the code is mostly from TuxAQFH - much cleaned up. Run it with
"./src/tuxfighter -n 2" for two players, "-n 4" for four players, etc
the control characters are displayed to stdout on startup and can
be changed in data/players.dat - there is no way to use the numeric
or arrow pads or function keys yet - but they have the same problems
as the main keyboard).

----------------------------- Steve Baker -------------------------------
Mail : <sjb...@ai...>   WorkMail: <sj...@li...>
URLs : http://www.sjbaker.org
       http://plib.sf.net http://tuxaqfh.sf.net http://tuxkart.sf.net
       http://prettypoly.sf.net http://freeglut.sf.net
       http://toobular.sf.net   http://lodestone.sf.net

Re: [Plib-users] Don't Do This!

[Christian M. <ma...@Ch...>]

Steve Baker wrote:
> 
> I puzzled about what might be wrong with my code - but eventually
> (after some questions to the LinuxGameDev mailing list and a bit of
> web surfing) discovered that this is a hardware limitation of the PC
> keyboard!
> 
> Aaaaggghhhh!!!!
> 
> I hadn't noticed it when testing because with just me pushing one
> or two keys, I never hit the combinations that cause problems.
> 
> This suprised me a lot (which is why I'm writing this email - so that
> others don't fall into the same trap).
> 
> The rule is approximately that if you hold down three keys that form
> the corners of a rectangle on the keyboards 'scan matrix' (E, C and I
> for example) then depending on the age of your keyboard, you'll either
> find that the third key you press will be ignored *or* the key at the
> fourth corner of that rectangle will appear to be pressed even though
> it isn't!
> [...]

Have you tried a USB keyboard? I could imagine that they aren't using
the old style method. And if that doesn'T work it's possible to try 2
USB or 1 USB and 1 PS/2 keyboard, etc.

CU,
Christian

--
The idea is to die young as late as possible.        -- Ashley Montague

Whoever that is/was; (c) by Douglas Adams would have been better...

Re: [Plib-users] Don't Do This!

[Steve B. <sjb...@ai...>]

Christian Mayer wrote:

> Have you tried a USB keyboard? I could imagine that they aren't using
> the old style method. And if that doesn'T work it's possible to try 2
> USB or 1 USB and 1 PS/2 keyboard, etc.

I don't think they'll be any different from the regular kind.

The problem isn't in how the data is transmitted from the keyboard to the
PC (that's basically just a serial protocol).  The problem is that the
decoder chip in the keyboard has only a limited number of pins (not one
per key) so it has to scan the keyboard using a row-column setup.

When you have two keys held down, two row lines are connected to two
column lines.  When you press down a third key that shared the same row
as one of those keys and the same column as the other, there is no way
for the keyboard controller chip to know which of the two keys were pressed.

Since the controller doesn't know which you pressed, it sends neither of
them (unless it's a really old keyboard controller chip - in which case
it sends *both*...Eeeek!)

I wrote up an Omniverous Biped FAQ item about this:

  http://www.sjbaker.org/steve/omniv/index.html

...check out "Keyboards Are Evil".


----------------------------- Steve Baker -------------------------------
Mail : <sjb...@ai...>   WorkMail: <sj...@li...>
URLs : http://www.sjbaker.org
       http://plib.sf.net http://tuxaqfh.sf.net http://tuxkart.sf.net
       http://prettypoly.sf.net http://freeglut.sf.net
       http://toobular.sf.net   http://lodestone.sf.net

Re: [Plib-users] Don't Do This!

[<ha...@sl...>]

>I wrote up an Omniverous Biped FAQ item about this:
>
>  http://www.sjbaker.org/steve/omniv/index.html
>
>...check out "Keyboards Are Evil".

Now, not to disagree with what you've figured out, as I'm sure it's
correct, but...  Whats weird is I've seen several older windows games that
did exactly this, with two players.  One in particular was action based,
kinda like a player versus player asteroids game.  I wish I could
remember the name.  You picked a space ship (each had different
speeds/weapons/etc).  You had to thrust, rotate and fire - all sharing
the same keyboard.

--
Brian

Re: [Plib-users] Don't Do This!

[Steve B. <sjb...@ai...>]

ha...@sl... wrote:
> 
> >I wrote up an Omniverous Biped FAQ item about this:
> >
> >  http://www.sjbaker.org/steve/omniv/index.html
> >
> >...check out "Keyboards Are Evil".
> 
> Now, not to disagree with what you've figured out, as I'm sure it's
> correct, but...  Whats weird is I've seen several older windows games that
> did exactly this, with two players.  One in particular was action based,
> kinda like a player versus player asteroids game.  I wish I could
> remember the name.  You picked a space ship (each had different
> speeds/weapons/etc).  You had to thrust, rotate and fire - all sharing
> the same keyboard.

I didn't say it was impossible - if you pick the right keys, it can work.

(eg Q,W,E,R as Turnleft/Thrust/TurnRight/Fire for player 1, 
    Y,U,I,O as the same for player 2...that works just fine).

Also, if you make the game so that a quick tap on the keyboard is all you
need then it's *unlikely* (but not impossible) for something to go wrong.

----------------------------- Steve Baker -------------------------------
Mail : <sjb...@ai...>   WorkMail: <sj...@li...>
URLs : http://www.sjbaker.org
       http://plib.sf.net http://tuxaqfh.sf.net http://tuxkart.sf.net
       http://prettypoly.sf.net http://freeglut.sf.net
       http://toobular.sf.net   http://lodestone.sf.net

Re: [Plib-users] Don't Do This!

[Christian M. <ma...@Ch...>]

Steve Baker wrote:
> 
> ha...@sl... wrote:
> >
> > >I wrote up an Omniverous Biped FAQ item about this:
> > >
> > >  http://www.sjbaker.org/steve/omniv/index.html
> > >
> > >...check out "Keyboards Are Evil".
> >
> > Now, not to disagree with what you've figured out, as I'm sure it's
> > correct, but...  Whats weird is I've seen several older windows games that
> > did exactly this, with two players.  One in particular was action based,
> > kinda like a player versus player asteroids game.  I wish I could
> > remember the name.  You picked a space ship (each had different
> > speeds/weapons/etc).  You had to thrust, rotate and fire - all sharing
> > the same keyboard.
> 
> I didn't say it was impossible - if you pick the right keys, it can work.
> 
> (eg Q,W,E,R as Turnleft/Thrust/TurnRight/Fire for player 1,
>     Y,U,I,O as the same for player 2...that works just fine).

Does the num pad count extra? If that's so than two players should be
possible.

CU,
Christian

--
The idea is to die young as late as possible.        -- Ashley Montague

Whoever that is/was; (c) by Douglas Adams would have been better...

Re: [Plib-users] Don't Do This!

[Steve B. <sjb...@ai...>]

Christian Mayer wrote:
> 
> Steve Baker wrote:
> >
> > ha...@sl... wrote:
> > >
> > > >I wrote up an Omniverous Biped FAQ item about this:
> > > >
> > > >  http://www.sjbaker.org/steve/omniv/index.html
> > > >
> > > >...check out "Keyboards Are Evil".
> > >
> > > Now, not to disagree with what you've figured out, as I'm sure it's
> > > correct, but...  Whats weird is I've seen several older windows games that
> > > did exactly this, with two players.  One in particular was action based,
> > > kinda like a player versus player asteroids game.  I wish I could
> > > remember the name.  You picked a space ship (each had different
> > > speeds/weapons/etc).  You had to thrust, rotate and fire - all sharing
> > > the same keyboard.
> >
> > I didn't say it was impossible - if you pick the right keys, it can work.
> >
> > (eg Q,W,E,R as Turnleft/Thrust/TurnRight/Fire for player 1,
> >     Y,U,I,O as the same for player 2...that works just fine).
> 
> Does the num pad count extra? If that's so than two players should be
> possible.

If you look at my FAQ:

  http://www.sjbaker.org/steve/omniv/index.html

You'll see a table at the bottom of the page showing (for my keyboards
at least) what the decoder's matrix looks like.

It predicts that holding down (for example) 'R' and 'T' will
block 4,5,6,7,8,9 and + on the keypad...<tries it>...and it does!

Really, no keys are immune to the effect.  You just have to *carefully*
pick keys that either all occupy different rows - or all occupy different
columns.

So, looking at that table in the FAQ, you'll see that there are 8 rows
and 18 columns.

Ack! I accidentally typed my table in 'sideways'...and I'm *certainly*
not going to type it in again. Tables are a pain in HTML!.


   Conclusion: You can have at most 18 keys active in the game.


However, it gets worse.  

Some columns of the matrix only have wierd keys like Left-Meta or Right-Meta
that are something of a pain to read in most window libraries...and
which can have unintended consequences.  (You wouldn't want to write
a game that could result in player 1 pressing Ctrl, Player 2 pressing
Alt and Player 3 pressing Del or Backspace in the heat of the action!!)

So, if we eliminate the columns containing only 'dangerous' keys,
we have 14 or maybe 15 keys left (depending on what 'PrintScreen' does).

That's certainly not enough for a four player mode for my game - and
even in 3 player mode, one of the players has to uses SysRequest/PrintScreen
as one of his keys in order to have 5 keys each - and I really wanted
at least six keys per player (North,South,East,West,Jump,Fire) and had
really hoped for something like 10 or 15 per player.

So, I'm left with two player mode as the only possible setup and only
seven keys each.

I can do a *little* better by knowing that you won't often want to press
North and South together - or East and West together...but still, I should
try to ensure that when a player does that by mistake, it doesn't mess up
his competitor.

There would also be *easy* scope for covert cheating!  Just hold down the
right key combination to block the other player's Fire button!  The game
could actually detect that and flash up a "CHEAT!" warning.

Another way to increase the number of commands per player would be to
write the code in a way that would discourage holding some of the keys
down for a long time.  For example, if holding down the 'Jump' button
for more than some small amount of time caused the player to trip and
fall flat on his face, then I could use more than one key in that column,
accepting the small probability of a short-term key lock-out.

There is also the issue of i18n.  There are at least a dozen keyboard
layouts in the world - and some (such as the 'AZERTY' design which has
A and Z where most people would expect Q and W to be) will need different
key assignments...then there are people who remapped their keyboards and
swapped the keycaps around to do a Dvorjak layout or something.

Most worrying of all is that I've only tested this out on the four keyboards
I have at home (Casio laptop, Fujitsu laptop and two different no-name
ergonomic 'split' keyboards).  There is no reason at all why some other
brand might not have a different keyboard decoder with a different key
matrix.  The PC and it's software wouldn't care - but the combinations
of keys that would lock out could be totally different.

----------------------------- Steve Baker -------------------------------
Mail : <sjb...@ai...>   WorkMail: <sj...@li...>
URLs : http://www.sjbaker.org
       http://plib.sf.net http://tuxaqfh.sf.net http://tuxkart.sf.net
       http://prettypoly.sf.net http://freeglut.sf.net
       http://toobular.sf.net   http://lodestone.sf.net

Re: [Plib-users] Don't Do This!

[Cameron M. <li...@to...>]

* sjb...@ai... (Steve Baker) [2001.12.30 21:32]:
> The rule is approximately that if you hold down three keys that form
> the corners of a rectangle on the keyboards 'scan matrix' (E, C and I
> for example) then depending on the age of your keyboard, you'll either
> find that the third key you press will be ignored *or* the key at the
> fourth corner of that rectangle will appear to be pressed even though
> it isn't!

Hmm...that is weird.

> I thought for a while about picking keys that didn't form the corners
> of such squares (eg E, C and G work at the same time) - but rapidly
> realised that I couldn't come up with anything remotely ergonomic
> in that way...not even for two players - let alone three or four!
> 
> So now I have to rethink this in terms of two players using joysticks,
> one using the keyboard and another using the mouse.

You want four people using the same keyboard?  Have you had too much
egg nog?  :-)  Seriously though, I don't think I've ever played a PC
game where more than two people used the keyboard at the same time, and
it seems impracticle to think people would want to in the first place.

Anyway, I wasn't aware of the goofy PC keyboard behavior you are brought
up.  Thanks for bringing it up...
-- 
Cameron Moore

Re: [Plib-users] Don't Do This!

[Steve B. <sjb...@ai...>]

Cameron Moore wrote:
> 
> * sjb...@ai... (Steve Baker) [2001.12.30 21:32]:
> > The rule is approximately that if you hold down three keys that form
> > the corners of a rectangle on the keyboards 'scan matrix' (E, C and I
> > for example) then depending on the age of your keyboard, you'll either
> > find that the third key you press will be ignored *or* the key at the
> > fourth corner of that rectangle will appear to be pressed even though
> > it isn't!
> 
> Hmm...that is weird.

Yes - exactly.

> You want four people using the same keyboard?  Have you had too much
> egg nog?  :-)

Well, no - but once you've written it for two players, it's only two
more lines of code to allow three or four.

>  Seriously though, I don't think I've ever played a PC
> game where more than two people used the keyboard at the same time...

It emerges that there is a *reason* for that!  :-(

> and it seems impracticle to think people would want to in the first place.

You'd be suprised what kids will do.  We've played glTron that way - it's a blast!

> Anyway, I wasn't aware of the goofy PC keyboard behavior you are brought
> up.  Thanks for bringing it up...

I didn't want anyone else to suffer that amount of pain!

----------------------------- Steve Baker -------------------------------
Mail : <sjb...@ai...>   WorkMail: <sj...@li...>
URLs : http://www.sjbaker.org
       http://plib.sf.net http://tuxaqfh.sf.net http://tuxkart.sf.net
       http://prettypoly.sf.net http://freeglut.sf.net
       http://toobular.sf.net   http://lodestone.sf.net

Re: [Plib-users] Don't Do This!

[Adam S. <ada...@ho...>]

Hello All -

> I've got to the point of dividing the screen into two, three or four=20
> chunks and rendering four cute little penguins - each of which can move
> independantly with a camera that follows along.  I had all of it workin=
g
> with the controls for each player assigned from a config file (eg=20
>=20
>   Player 1: MoveNorth =3D W, MoveSouth =3D Z, MoveEast =3D W, MoveWest =
=3D S
>             Jump =3D Q, Fire =3D E
>   Player 2: MoveNorth =3D I, MoveSouth =3D M, MoveEast =3D J, MoveWest =
=3D K
>             Jump =3D U, Fire =3D O

	I've been thinking about this for a while, and as far as I can
tell, this should work in all situations, except if a person tries to go
north, south, and east or west at the same time, which shouldn't happen
unless a player is really discombobulated:

  Player 1: MoveNorth =3D 2, MoveSouth =3D W, MoveEast =3D E, MoveWest =3D=
 Q
            Jump =3D 5, Fire =3D G
  Player 2: MoveNorth =3D 0, MoveSouth =3D P, MoveEast =3D [, MoveWest =3D=
 O
            Jump =3D 8, Fire =3D M

	Every once in a while, my fingers get a super power, and I'm
able to hit <north or south>, <west or east>, Jump, and Fire for both
players and it works!  :-D

	Thanks for pointing that out!


--=20
    Real programmers don't comment their code.  It was hard to write, it
    should be hard to understand.

   /||  Adam Seyfarth    <http://members.home.net/adam.seyfarth/>  ||\   =
/"\
  /=AB||  <cl...@us...>        ASCII Ribbon, HTML-free email  ||=BB=
\  \ /
  \=AB||  GU d-- s+: a---- C++ UL+>+++ P-- L++ E-- W++ N++ o K- w--  ||=BB=
/   X
   \||  O M- V- PE- Y PGP- t 5 X R  tv+ b+ DI+ D+ G e-- h! !r !y+  ||/   =
/ \

Re: [Plib-users] Don't Do This!

[Steve B. <sjb...@ai...>]

Adam Seyfarth wrote:

>         I've been thinking about this for a while, and as far as I can
> tell, this should work in all situations, except if a person tries to go
> north, south, and east or west at the same time, which shouldn't happen
> unless a player is really discombobulated:
> 
>   Player 1: MoveNorth = 2, MoveSouth = W, MoveEast = E, MoveWest = Q
>             Jump = 5, Fire = G
>   Player 2: MoveNorth = 0, MoveSouth = P, MoveEast = [, MoveWest = O
>             Jump = 8, Fire = M

Yes - that's not a bad set - I've been using:

   Player 1: MoveNorth = 2, MoveSouth = W, MoveEast = 3, MoveWest = 1
             Jump = Q, Fire = E  (With other moves mapped to 4/5/R/T)
   Player 2: MoveNorth = J, MoveSouth = M, MoveEast = K, MoveWest = H
             Jump = N, Fire = <  (With other moves mapped to L/;/>/?

...and I've given up on the idea of four players.

Because one player is confined to the top two rows and the other to the
bottom two, they can't interfere with each other.  They *can* lock themselves
out of some keys (eg Jumping or shooting whilst moving South-East or South-West
is bad).

I think your combo avoids that - but it hurts my head to try to prove it!

I have one of those sexy split keyboards (try one if you havn't - you'll
never go back once you're used to it) - and this mapping works out nicely
with each player confined to two rows and all the columns on one side of
the split.

The *ultra* bad news is that I have definite evidence that some PC keyboards
are scanned differently from others so there probably isn't a bullet-proof
combination that'll work for everyone.  One of our no-name keyboards at
work has Y,H,N and T,G,B on the same column scan and another has Y,H,N
sharing with Q,A,Z - so I think I'm basically *doomed*.

I'm just going to document what I know and make the keyboard controls
completely programmable.

----------------------------- Steve Baker -------------------------------
Mail : <sjb...@ai...>   WorkMail: <sj...@li...>
URLs : http://www.sjbaker.org
       http://plib.sf.net http://tuxaqfh.sf.net http://tuxkart.sf.net
       http://prettypoly.sf.net http://freeglut.sf.net
       http://toobular.sf.net   http://lodestone.sf.net