[gclist] Another request from a Newbie :-)
Mon, 19 May 2003 11:49:43 +0200
> > There are even cases of GC techniques that claim to be real-time
> > since the shortest measured interrupt time is X. This is simply not
> > enough for hard real-time.
> I disagree. Many (I might even say most) real-time applications are
> developed using exactly that kind of measure, often with a specific
> (maybe highly demanding) test suite. Some real-time applications
> (e.g. safety-critical systems) may have a more rigorous method for
> ensuring that real-time constraints are met, but many, many
> applications do not.
I think we agree on everything here but the definition of hard real-time. It
is common to define hard real-time systems as systems that must not miss any
deadline, and soft real-time systems are systems where the quality of the
output decrease if deadlines are missed (like video decoders etc.) Using
this "definition" of hard real-time, it is simply not enough to claim that
the system worked fine in tests to lauch them.
Since the memory requirements of dynamically garbage collected systems are
really hard to predict, one should be very careful to use dynamic GC in
these systems. However, if the memory requirements are known, there are
several techniques to choose from. Unfortunately, they all consume a lot of
memory, but that is another thread ;) And again, I wonder if knowing the
memory requirements is the same as knowing when objects turn into garbage in
practice. (Which should imply that the system does not need GC anymore.) Has
anybody looked into that?