Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!ames!haven!umd5!jonnyg
From: jonnyg@umd5.umd.edu (Jon Greenblatt)
Newsgroups: comp.windows.ms
Subject: Re: C_Talk review and final review if C
Message-ID: <5270@umd5.umd.edu>
Date: 25 Aug 89 15:23:16 GMT
References: <5258@umd5.umd.edu> <237700003@ENR.Prime.COM>
Reply-To: jonnyg@umd5.umd.edu (Jon Greenblatt)
Organization: University of Maryland, College Park
Lines: 53

In article <237700003@ENR.Prime.COM> LANDON@ENR.Prime.COM writes:
>
>/* Written  4:24 pm  Aug 24, 1989 by jonnyg@umd5.UUCP in ENR:comp.windows.ms */
>/* ---------- "C_Talk review and final review if C" ---------- */
>
>>First of all I finished reviewing CommonView C++. I found it to
>>be mostly broken.
>
>You better come clean, because this statement has no credibility.  I've
>been working with CommonView since its release, and while I've had my
>frustrations, it's been due to "user error" in most cases.  There are bugs

[ REMOVE FOOT FROM MOUTH... ]

	I guess I should come clean. I never said the compiler was broken.
The compiler is great. My frustrations have been with the fact I need a
certian minimum functionality and CommonView is too new to supply it
without writing it myself. The memory management hacks do work nicely into
media independent databases but I would still not like to be painfuly
reminded about this when I try to use them, maybe getting used to them would
help. I feel bad that I have shown too much bias against Glockenspiel C++.
Like I say the compiler is nice! The class library lacks what I need as
a windows programmer but maybe I am biased by the types of apps I need to
write. The work I do is best done in a highly virtual environment with
garbage collection. If lisp were an assembly language instead of an
interpreter I guess I would have it made.

	In summary, please excuse me for venting my frustrations on the net.
I knew if I were realy off base someone would correct me. What I need
now is a fast OOPS with garbage collection, virtual operators, and a decent
syntax. Maybe I should take up compiler writing! Switching heaps, and
object construction/destruction based on scope are good Ideas, I guess
I want more! I think it would be nice if a compiled language could keep
tract of all pointers and be able to traverse links based on data type, a
garbage collection scheem could then be worked into this. I guess I can
dream at least....

	C_Talk has it's own version of overloading. It's a hack that requires
you to name each parameter as you pass it to the message dispatcher. I prefer
the C++ method of function/operator overloading. The C_Talk environment is
fairly straight forward and easy to use. There are a lot of benifits to
this but there are also losses due to this simplicity. The class library
in C_Talk is based around windows and has been from the start. I am
still frustrated however. I have been dealing with interpreted environments
for some time now (Actor/Lisp-OOPS) and its hard to get back into the
restrictions compilers force on you.

	Any compiler writers out there willing to work with me on this?
Maybe I could eek up my Idea environment using yacc, then I would show
you all!!! Aren't nets nice, you can make a fool out of youself and
you can almost fool yourself into thinking no body even notices.

						JonnyG.