Re: tracing user actions

[email protected] wrote:

>Hi,
>I'm soon to be involved in user-computer interactions studies.
>We will need to collect usage data to analyze how students are using a software environement.
>I would like to be able to use the gnome software, but I would need to be able to log *every* user interaction with applications. Not on the X events side, but on the applications side. 
>I thought that libs could be hacked so that upon processing a gnome (gtk?) event, the processing object could write some logging information.
>I have zero gnome-gtk knowledge for now, and would like to know if this would be possible at the gnome level, or if I have to used an environement OO toolkit. (No flame intended but GUI elements could be inheriting from an EventObserver class with a virtual member logging function querying for the precise GUI element)
>
>I've been searching for "tracing" and "logging" on the mailing lists archives, but to no avail ...
>GtkTrace might have been useful, but it disapeared.
>
>I am willing to code, and think that both code and results of the study could be very useful to Gnome.
>But of course I would like to write the minimum code (that's why I'm lloking for a per-library approach instead of a per application approach), and I would also like to avoid rewriting something already available.
>
>Any ideas ?
>
>Thank you very much in advance.
>  
>
In GTK 2.0 and GNOME 2.0 a lot of work has been put into accessibility. 
 A side effect of this is that you can introspect a lot of information 
about the user interface through the ATK (in process) or AT-SPI (out of 
process) interfaces.  This information is probably closer to the level 
you are after (what user interface element has focus, what actions a 
user interface element can perform, etc), compared than straight X 
events.  This might be exactly what you are looking for.

James.

-- 
Email: [email protected]              | Linux.conf.au   http://linux.conf.au/
WWW:   http://www.daa.com.au/~james/ | Jan 22-25   Perth, Western Australia.