Newsgroups: comp.object,comp.lang.smalltalk,comp.lang.ada
Path: cantaloupe.srv.cs.cmu.edu!rochester!udel!news.mathworks.com!news.kei.com!world!bobduff
From: bobduff@world.std.com (Robert A Duff)
Subject: Re: Has C++ had its day?
Message-ID: <DAJ73H.35M@world.std.com>
Organization: The World Public Access UNIX, Brookline, MA
References: <3q7tb1$1ct@holly.csv.warwick.ac.uk> <1995Jun12.135705.3580@merlin.hgc.edu> <3rsnp8$bqj@wally2.hti.net> <3s7ijs$jjb@gensym1.gensym.com>
Date: Wed, 21 Jun 1995 16:28:29 GMT
Lines: 12
Xref: glinda.oz.cs.cmu.edu comp.object:33302 comp.lang.smalltalk:24873 comp.lang.ada:31575

In article <3s7ijs$jjb@gensym1.gensym.com>,  <jsa@gensym.com> wrote:
>Now this seems to contradict your earlier statement.  For example, any
>_additions_ to the spec couldn't modify any of the dependencies and thus
>would not require any recompilation.

Not true.  If you add a function that overloads something else (possibly
in some faraway package), then this could affect overload resolution of
something that can see both.  A previously-legal call to Foo could
become ambiguous, because you added a new thing called Foo.  It's not
trivial for an incremental compiler to get this right.

- Bob
