[gcompris-devel] Log Module for gcompris

[Ralf Jardon]

Am Do, den 05.02.2004 schrieb Bruno Coudoin um 22:55:
> Le mer 04/02/2004 à 12:54, Ralf Jardon a écrit :
> > Am Mi, den 04.02.2004 schrieb Bruno Coudoin um 01:27:
> > > I just commited the implementation of the trace format that includes the
> > > keylog. 
> > 
> > Wow...
> > 
> > 
> > > Ralf, it's up to you now to tell us if this is practical.
> > 
> > OK, first try:
> > 
> > gcompris_log_set_key k/6:k/1:k/2:q/1:g/5:g/0:g/1:f/2:x/2:f/7:f/1:f/0:t/1
> > 
> > The keylog is good but we have forgotten something :-)
> > With this log i am not able to see the correct key.
> > I have pressed the "k" three times but the "q" was the solution (the
> > only visible letter on screen). 
> > I cannot reconstruct the task. But the task is necessary to analyse the
> > faults of the children.
> > 
> > Perhaps we can solve the problem in this way:
> > 
> > gcompris_log_set_key
> > k/6:k/1:k/2:(q)/1:g/5:g/0:g/1:(f)/2:x/2:f/7:f/1:f/0:(tzg)[...]
> > 
> > The letters in brackets are the correct (visible) letters, if more then
> > one letter are on screen, we can log all visible letters e.g. (tzg)...
> > 
> 
> I believe it's becoming overly complex. Rereading the wiki at
> http://www.ofset.org/wiki/index.php/GcomprisLogFormat
> I think we can achieve what you expect without adding the keylog.
> In this case, we could set a fail log each time he/she does wrong in the
> comment field. You could get someting explicit like:
> 2004;bruno.maison;bcoudoin;gcompris;gletter;1;2;FAILED;14;The user key
> is 'a' the choices were 'cde'

It looks very good. All informations i need are in the line. The only
unpractically is the missing seperator between 'key' and 'choice'
(needed for CSV import). So i suggest:

[...]14;The user key is 'a'; the choices were 'cde'

Is it possible to implement it untill next Monday?

The study began and I need the program starting from Thuesday in the
Kindergarten. 

The evaluation of Gcompris is part of my thesis. So - hopefully - you
will get a scientific effectiveness proof in a few months. In the
meantime i'll be glad to share some major results with you. I hope, with
this results and with consideration of psychological theories, we are
able to furnish proof of effectiveness. Based on this major proof we can
improve the effectiveness of gcompris gradually.

Next time i'll tell you something about the theory of 'inductive
thinking'. The 'babymatch' board is an example of this theory of mind,
but we have to extend it a little bit in order to be able to work
thereby scientifically.

Hope it's ok for you ;-)

> I can't imagine a teacher being able to read the keylog as we have it
> now, too cryptic.

Agree...

Ralf.