← Back to team overview

multi-touch-dev team mailing list archive

Re: Testing M-t without HW

 

Wrote a small proof of concept bash script to test replaying an Xnee sessions file to swinput (found an extra mouse that I attached to get multiple input devices).

The bash script wrote instructions for multiple input devices down to the kernel. Really (I mean it) not a nice script, but sure thing, it works. The kernel "sends" relative motions to X. Need to turn them relative motions into absolute.

As far as the other (really he was the first!) Henrik goes, I believe he lives in the same town as I do (Gothenburg, Sweden)?

/h





On 09/17/2010 09:25 PM, Duncan McGreggor wrote:
Hello, Henrik -- thanks for introducing yourself! New humans are always
welcome :-)

Your email is well-timed. We're in the process of defining ways in which
we can deliver testing for uTouch. Another Henrik on the list is
currently working on getting latency info out of various layers of the
stack. We'll be having discussions about this at UDS (with live, remote
participation) and it's one of the big areas we're focused on for Natty.

Great thoughts and ideas -- keep them coming! And we'll be checking
these out...

d

On 09/17/2010 12:36 PM, Henrik Sandklef wrote:
hi all

  I am new human on this M-t group so I may step one someone toes with
the following suggestion.  While reading a bit on M-t I got two ideas
for testing M-t, without hw:

1. Xnee
---------------
  Xnee can record and replay multiple input devices, although it needs
some more work.

  A developer can record a set of sessions using Xnee. These session
typically include gestures that need to be tested. These sessions can be
replayed during regression test, to verify an application, a lib or
unity at a computer without M-t input device ...


2. Xnee + swinput
---------------
  Record a session as above. But instead of replaying it on the X level,
swinput can fake hw input devices.


So the basic idea is that person with M-t hardware records a couple of
sessions which can be replayed at any computer.


DISCLAIMER: I am aware that scripted sessions are not going to solve all
problems, but it may solve (hmmm, and introduce) some. And as always,
automated tests smells a bit like snake oil ;)



regards, henrik




Xnee
==================================
The basic features of Xnee:

* Records X11 data using the RECORD extension
* Replays device events using XTest
    syncs replay using all data but device events

I've just added support for XInput (not XI2, but rather "XI 1.5") to Xnee.


Swinput
==================================
Fakes 1-xx HW input devices via the kernel. So no X11 faking at all.

Swinput has had the support for a while.


swinput mice as seen by xinput:
----------------------------------

⎡ Virtual core pointer       id=2    [master pointer  (3)]
.........
⎜   ↳ swinput faked mouse device id=14    [slave  pointer  (2)]
⎜   ↳ swinput faked mouse device id=15    [slave  pointer  (2)]

_______________________________________________
Mailing list: https://launchpad.net/~multi-touch-dev
Post to     : multi-touch-dev@xxxxxxxxxxxxxxxxxxx
Unsubscribe : https://launchpad.net/~multi-touch-dev
More help   : https://help.launchpad.net/ListHelp





Follow ups

References