[9fans] rio /dev/mouse

[9fans] rio /dev/mouse

[YAMANASHI Takeshi]

I'm using Microsoft Intelli wheel mouse now.

Without rio, I rolled up and down the wheel while doing
`cat /dev/mouse', and its button state of the output was
`8' or `16'.

But under rio, `cat /dev/mouse' produced with the button
state `0' only.

Could you tell me what is happening and why?

I want to change rio to interpret this 8 or 16 button state
as either arrow-up or arrow-down keyboard event.
-- 
Sincerely,
YAMANASHI Takeshi

Re: [9fans] rio /dev/mouse

[Boyd Roberts]

YAMANASHI Takeshi wrote:
> 
> I'm using Microsoft Intelli wheel mouse now.

I have a Logitech Wingman Gaming Mouse on a serial interface.

aux/mouse does not like it.  It's bound to be some stupid
dance that has to be done to get the mouse to behave.  If
I had an RS-232 protocol analyser I'd fix it.  Hacking
aux/mouse looks like no fun, because it has to deal
with all that 'orrible RS-232 glop.

Re: [9fans] rio /dev/mouse

[YAMANASHI Takeshi]

> :  And of course, I have changed rio/rio.c so that rio is now
> :  interpreting rolling wheel as either Kup or Kdown pressed.
> 
> Can you send the code?

Sure.  I only added the following lines headed with `>' to rio.c. 
I don't know this works in general, however.

in mousethread() in rio.c:

	for(;;)
	    switch(alt(alts)){
		case MReshape:
			resized();
			break;
		case MMouse:
>		if (mouse->buttons == 8) {
>			static Rune up[2] = {Kup, L'\0'};
>
>			if (input != nil)
>				sendp(input->ck, up);
>			continue;
>		}
>		else if (mouse->buttons == 16) {
>			Rune down[2] = {Kdown, L'\0'};
>
>			if (input != nil)
>				sendp(input->ck, down);
>			continue;
>		}
		Again:
--
YAMANASHI akeshi

Re: [9fans] rio /dev/mouse

[Boyd Roberts]

> I changed the line
>       m.buttons &= 7;
> to
>       m.buttons &= 127;

s/127/0x7F/

it's far more obvious.

Re: [9fans] rio /dev/mouse

[Fco.J.Ballesteros]

:  I changed the line
:  	m.buttons &= 7;
:  to
:  	m.buttons &= 127;

Don't have that line, perhaps I just changed it and then forgot I changed it.

:  And of course, I have changed rio/rio.c so that rio is now
:  interpreting rolling wheel as either Kup or Kdown pressed.

Can you send the code?

Re: [9fans] rio /dev/mouse

[Russ Cox]

> High bits of "m.buttons" is cleared in libdraw/mouse.c.
> I changed the line
> 	m.buttons &= 7;
> to
> 	m.buttons &= 127;

Have you applied the 10140327 full update from
the Oct 14 2000 release to the March 27 2001 release?
We dropped that line completely between the two.

What does wrap/info plan9 print?

Russ

Re: [9fans] rio /dev/mouse

[YAMANASHI Takeshi]

> Have you applied the 10140327 full update from
> the Oct 14 2000 release to the March 27 2001 release?
> We dropped that line completely between the two.

I am ashamed to say the truth, but no, I haven't any yet.
I was not sure in what order I should apply updates.

> What does wrap/info plan9 print?

	plan9 (full package as of Wed Jan 31 19:22:50 JST 2001)
	pkgupd 980936570 updating 971556349: Plan 9 - wiki software
	pkg 971556349: Plan 9 from Bell Labs ? Third Edition
-- 
Sincerely,
YAMANASHI Takeshi

Re: [9fans] rio /dev/mouse

[YAMANASHI Takeshi]

Thank you for your enlightenment, rob.

> Long ago, various pieces of the system cleared out the top bits but it
> looks to me like all that code is gone.

High bits of "m.buttons" is cleared in libdraw/mouse.c.
I changed the line
	m.buttons &= 7;
to
	m.buttons &= 127;

And of course, I have changed rio/rio.c so that rio is now
interpreting rolling wheel as either Kup or Kdown pressed.
-- 
Sincerely,
YAMANASHI Takeshi

Re: [9fans] rio /dev/mouse

[rob pike]

[-- Attachment #1: Type: text/plain, Size: 540 bytes --]

Long ago, various pieces of the system cleared out the top bits but it
looks to me like all that code is gone.  A little debugging should
tell you where the problem is.  Put a print into
/sys/src/cmd/rio/rio.c, in the routine mousethread(), right after case
MMouse.  Print out the state of mouse.buttons.  If the button is
cleared there, it's the library (/sys/src/libdraw/mouse.c); if it's
still set here, then rio is doing it somewhere.  I don't have an
intellimouse handy but this should be easy for you to track down.

-rob


[-- Attachment #2: message.txt --]
[-- Type: message/rfc822, Size: 2084 bytes --]

From: YAMANASHI Takeshi <uncover@beat.cc.titech.ac.jp>
To: 9fans@cse.psu.edu
Subject: [9fans] rio /dev/mouse
Date: Fri, 22 Feb 2002 14:35:22 0900
Message-ID: <200202220550.g1M5oJk246675@mail-o.cc.titech.ac.jp>

I'm using Microsoft Intelli wheel mouse now.

Without rio, I rolled up and down the wheel while doing
`cat /dev/mouse', and its button state of the output was
`8' or `16'.

But under rio, `cat /dev/mouse' produced with the button
state `0' only.

Could you tell me what is happening and why?

I want to change rio to interpret this 8 or 16 button state
as either arrow-up or arrow-down keyboard event.
-- 
Sincerely,
YAMANASHI Takeshi