To be live or not to be live!
Jecel Assumpcao Jr
jecel at merlintec.com
Tue Apr 23 15:39:22 UTC 2002
On Monday 22 April 2002 20:27, Bruce Cohen wrote:
> One of the reasons the Smalltalk-based oscilloscopes made by
> Tektronix were such a successful product was *because* you could be
> confident there were no memory leaks.
Well, you can have memory leaks in Smalltalk just by having objects you
don't expect (undo buffers and other stuff) holding on forever objects
you think were eliminated. But this is serious enough that we make the
effort to deal with it most of the times people come across this
problem.
We don't have dangling pointers, which are even nastier.
> Electronic instruments like
> oscilloscopes are often used in long-running measurements (hours,
> days. weeks-long sometimes) where it can be very inconvenient, even
> downright disastrous to an experiment to have to restart the
> equipment.
The ideal living system shouldn't have to be restarted even while being
upgraded. I think we will get there, eventually.
> So not having to worry about the firmware memory (which
> is usually pretty small to start with) filling up and hosing the
> scope's sampling controller or user interface is a very important
> feature.
Yes, the "low space notifier" is likely to show up only during
development, not during normal use.
> I like to think that if it weren't for Ada (and you think Java is
> bad!), Smalltalk might have become the common language for avionics
> for this reason among others.
I am not sure that the hardware was up to it (though a custom 68030
board was no problem for those kinds of applications...).
-- Jecel
More information about the Squeak-dev
mailing list
|