Squeak: who's afraid of the big bad "crash"?
moroz at paragraph.ru
Tue Dec 15 12:57:46 UTC 1998
> I like to say the ability to willingly and intentionally
> crash a software
> environment is an important feature -- even if one would only
> use it for
> venting ("balance of power"). Moreover, fragility is the
> price we pay for
> the reflective (meta) capabilities.
When I see the development environment that can reliably detect the user's
intention to crash it and then crash only in this case, I will probably
agree with you. I may sound like a nonsensical idiot, used to program in C++
and Java much more than Smalltalk (at least that's what pays my hobbies ;)),
but until then I say firmly: "The development environment should not crash,
intentionally or not. Every crash is a bug."
Dangerous operations should at least ask the developer "Is this really what
you want to do ?" and offer some checkpointing. At "runtime" these should
probably lead to graceful system exit by default.
Also, I don't see any inherent fragility in reflection by itself. You can
make meta-facilities safe. You can "split" the system when the meta-level is
changing and provide a way to revert to the old behaviour. Etc, etc...
More information about the Squeak-dev