🚀 go-pugleaf

  To: comp.lang.clos@cis.ohio-state.edu
  Date: Thu, 20 Aug 1992 07:23:55 GMT
  Organization: The Ohio State University Department of Computer and Information Science
  From: john@linus.mitre.org
  Sender: commonloops-request@cis.ohio-state.edu
  Subject: Re: Beginner's question and method lookup
  
  Eric Smith (eric@tfs.com) asks:
  
    But doesn't that make method dispatch very time consuming?  It
    seems like it would involve searching for the method in tables, and
    that the amount of time consumed would depend on how many methods
    were in those tables and how many class arguments each method had.
  
  and David C. Martin (dcmartin@msi.com) replies:
  
    Yes, if you did things that way.  In fact, you can determine at
    compile time what the argument types are and then put in the
    appropriate call to the appropriate function.
  
  I think this is often true, but not in general, even with extensive
  declarations.  If you have a very rich class hierarchy, the compiler
  may only be able to narrow the type to partway down the hierarchy, or
  to a disjunction of types, not to a leaf.  A SSC may be able to make
  use of this information, but I'm not sure many commercial CLOSes could.
  
  In the most ridiculous case, an argument to a generic-function may be
  computed via a call to READ; there's little the compiler can do then.
  
READ is not a ridiculous case, it is a ridiculously important one.  It is
typical of any interaction with the user that could return an arbitrary
object.  While C++ may be clever about using compile time type information,
one of its short coming is it does not support run time typing adequately
enough, READ being a good example.

If there were sufficiently smart compilers, there would be no programmers.

k