Re: If GC is the solution, then what is the problem?

David Abrahams wrote:


It's not silly.

One could ask, "What percentage of real world problems have an inherent
need for software? Zero."

Balance in arugment is lacking here. That's what it's all about anyway
right? The engineer looks at his (or her) landscape, considers all
things, and makes judgement calls on the relative merits of each
artifact in that landscape.

What I have been saying, and I will probably never deviate from, is
that, all things considered, GC seems to me to be more abused than
necessarily used.

Note that I never said that it was entirely unnecessary. I have only
said that the problems that it attempts to "address" are often problems
that never would have existed in the first place if an alternative
design had been chosen, one that is cleaner, more regular,
maintainable, etc.

Graphs are often mentioned as a domain where GC might be useful.
Agreed. However, not less than 10 minutes ago I revisited code using my
Graph<> class and it contains no exposed pointers:

struct Topology_Graph : protected Graph<Textual_Address, Node,
Network_Interface::Identifier, Network_Interface::Link>
{
    using Graph::Path;

-Le Chaud Lapin-

      [ See http://www.gotw.ca/resources/clcm.htm for info about ]
      [ comp.lang.c++.moderated. First time posters: Do this! ]