[M3devel] [M3commit] CVS Update: cm3

[Daniel Alejandro Benavides D.]

Hi all:
it seems that by denying itself from being well behaved program is checking for the unmatched traces or good behaviour programs, sort of analyzing many times more (incorrect traces with correct one) than checking the traces are well behaved and just checking wheh its not, better to check when its not, and not asumming the traces are ill anyway, then most of the time is spend in good standing than in bad or ill behaved program nalysis. 
The question how do you get those traces in good state, for instance checking by itself several times you would gain a pattern is repeatable, opposite of and ill corrupt and unmatched execution traces.
In that case it makes a trace pattern you are in good stands but of course if it doens't then you have sort of a problem there. The next step is where to get the break point before it goes into an unrepeatable trace, that is when you must have an static analyzer which run a simulation and verifies program is well stated at each assert of it.
Then it seems like you were getting the opposite direction going to see where it is not working yet at each part it makes asserts it, sort of looking for not well behaved errors but rather irrepeatable ones, which I think is harder. A error one you get by denying it's safe guarded makes illegal steps in the program you state or forbidden ones, is an unsafe error, a safe error is if you get it to pass it but you can't get it works by its results (sort of visbile) so getting in the right results but not in its good states are plausible to find, but ill ones, sort of what you are looking for i.e. looking for forbidden states that are ill-behaved  and get wrong results are almost the harder path to get it such a bug 
Just to let you know if you see this point. 



--- El sáb, 5/2/11, Mika Nystrom <mika at elego.de> escribió:

> De: Mika Nystrom <mika at elego.de>
> Asunto: [M3commit] CVS Update: cm3
> Para: m3commit at elegosoft.com
> Fecha: sábado, 5 de febrero, 2011 18:13
> CVSROOT:    /usr/cvs
> Changes by:    mika at birch.   
> 11/02/05 23:13:26
> 
> Modified files:
>     cm3/m3-libs/m3core/tests/thread/src/:
> Main.m3 
> 
> Log message:
>     removed a lot of unnecessary heap memory
> allocation, which could interfere with the threading tests.
>     
>     Most output to Stdio.stdout, not sure if
> this is right (compared with Stdio.stderr).
> 
>