Topic: Safe C++


Author: lance@eco.twg.com (Lance Kibblewhite)
Date: 18 Nov 1994 14:56:21 GMT
Raw View

I remember reading some posts about work on creating a standard for a safe
version of C++.  Is anybody actively pursuing this?

-- Lance.





Author: ellis@parc.xerox.com (John Ellis)
Date: 19 Nov 1994 10:34:34 GMT
Raw View
Lance Kibblewhite writes:

    I remember reading some posts about work on creating a standard
    for a safe version of C++.  Is anybody actively pursuing this?

Dave Detlefs and I wrote a two-part paper about pointer-safe C++:

    Part 1 contains a proposal for adding garbage collection
    to C++.

    Part 2 contains a proposal for a pointer-safe pure subset of C++,
    showing that such a subset is no different than the safe subsets
    of Modula-3 or Ada.

Part 1 was published in the April '94 Usenix C++ proceedings.  The
full paper is available from ftp.parc.xerox.com:/pub/ellis/gc.

We have actively pushed part 1, our proposal for garbage collection.
We started too late to get such a radical idea into the current
emerging standard, but we hope that when vendors do start thinking
about GC, they'll start by reading our paper.  (We've talked with many
of the major vendors.)  Hans Boehm maintains and distributes a
high-quality conservative GC that is highly compatible with almost all
C++ implementations (see ftp.parc.xerox.com:/pub/gc).  There is a C++
base-class interface in the collector that approximates the our GC
proposal.

We never pushed part 2.   We designed a pointer-safe-subset as an
exercise in thinking about C++ GC and as an answer to academic
researchers who thought that using GC with a pointer-unsafe language
was an oxymoron.   Though we think a pointer-safe subset is entirely
practical, it is entirely foreign to most C++ programmers.   Getting
GC into C++ is much more important.