[IGSTK-Developers] State machine logic for private functions
Luis Ibanez
luis.ibanez at kitware.com
Thu May 25 15:52:39 EDT 2006
Hi Julien,
The fact that the private functions are going to be called by friends
doesn't make them totally safe.
This is because "Frienship" in C++ just means that your friends have
access to your private data, it doesn' mean that they have good
intentions, or that they are competetent, or qualified for making
decisions for you. C++ "Friends" should actually be called
"Confidents", because their role is limited to have access to your
private information.
It is still important to verify that those private methods will only
be called when the class is in a state where it makes sense to
execute those methods. (Even if the caller is a so-called "friend").
Can you please describe the specific case that you are dealing with ?
Thanks
Luis
-----
Each time that you think of making (programmers) life simpler,
you should put it in the context of the 84,000 people who died
annually in U.S. hospitals as a result of medical errors.
What amount of programmer's life simplicity can justify a crash
in a software that is being used in the middle of a surgery ?
How many patients can be put at risk in order to save a couple
of hours of programming time ?
I would say that our life will easier if we can actually sleep
with peace of mind after IGSTK has been used in a surgery.
==================================================================
Julien Jomier wrote:
> Luis (and other state machine gurus)
>
> Do we really need state machine logic for functions that are private in
> one class and accessible only by friend classes? I would assume that all
> the state machine logic will be performed in the friend class.
>
> I know that by design this would be better but I'm just wondering...
> (and trying to make life simpler ;))
>
> Let me know,
>
> Julien
>
> _______________________________________________
> IGSTK-Developers mailing list
> IGSTK-Developers at public.kitware.com
> http://public.kitware.com/cgi-bin/mailman/listinfo/igstk-developers
>
>
More information about the IGSTK-Developers
mailing list