Re: Garbage Collection - The Trash Begins To Pile Up
Timo Geusch wrote:
That very much depends on the point in the development cycle when the
programmer's mistake is discovered. I believe there are a few
(software) engineers out there who may be pointing at some very real
Ariane V fragments to counter you assumption.
There is a break here. I never said that bad software would not
"ultimately result" in material costs. I said that, in the
construction of software, if you forget to initialize a variable, or
something like that, your prototype does not explode, not in
hour-to-hour development. Your example for Ariane is not entirely
analogous, as you do not find aeronautical engineers going out to the
launch every 3 hours to launch a new rocket.
Your statement only holds true if you assume that all bad design
decisions or bad implementation decisions get discovered during
implementation and testing. The newsgroup comp.risks is full of
counter-examples that IMHO prove his not to be the case.
If course. If you have a bad design that manifests in the field, then
yes, there is a material cost in any case, software _or_ hardware. It
should have been obvious from my post that I was referring to the cost
of materials during development. Otherwise, almost any commercial
software could have "significant material costs". Even something like
a bug tool for pirating software of the Internet can have "significant
material costs" if it leaves open a back door.
I am talking about routine, day-to-day development, inside the
development organization.
I think you've partially fallen into the trap of seeing GC as a(nother)
way of "dumbing down" C++ and somehow making it safe to use for some
kind of assembly line programmer, a task for which C++ is extremely
unsuited. Adding optional GC to C++ simply gives you (as the writer of
well-designed systems) another tool in the already rather large C++
toolbox and as such is a laudable goal as long as it is _optional_.
And I say again, I would like the C++ committee to add a keyword that
lets me specify that arrays can be copied and passed by values to
functions:
char x[32], y[23];
x = y; // I want this.
Of course, C++ already has rules governing this, so I cannot have my
way unless they add some keywords, making it "optional".
-Le Chaud Lapin-
--
[ See http://www.gotw.ca/resources/clcm.htm for info about ]
[ comp.lang.c++.moderated. First time posters: Do this! ]