Path: utzoo!utgpu!jarvis.csri.toronto.edu!mailrus!cs.utexas.edu!swrinde!ucsd!ucbvax!hplabs!hpl-opus!hpnmdla!hpmwtd!jeffa
From: jeffa@hpmwtd.HP.COM (Jeff Aguilera)
Newsgroups: comp.lang.c++
Subject: Re: Some questions on Zortech C++
Message-ID: <1520012@hpmwjaa.HP.COM>
Date: 10 Jan 90 16:01:56 GMT
References: <1172@zip.eecs.umich.edu>
Organization: HP Microwave Tech. - Santa Rosa, Ca.
Lines: 39

> <Here are some questions on bug/feature of Zortech C++:
> <(1) overload &:
> <    friend String operator&(String&, String&);
> <    will cause compiler error. (found in compile libg++ String.cc).
> <    This is a bug.
> Fixed in 2.06.

When is 2.06 shipping?  The 2.0 compiler I received yesterday has this same
annoying bug.  (My temporary patch is to overload the comma operator, and
add extra parentheses to compensate for the lower precedence.  Ack!)

Zortech must be shipping a lot of C++ compilers.  The package I received
was labelled 3" media, and I had ordered 3-1/2" media, but the box was filled
with those damned 5-1/4" suckers, with the diskette package seal already 
broken.  I expected <iostream.h> and <complex.h>, but neither is included
in the product; and <stream.hpp> is badly broken, returning _eof one 
character too soon.  That's right,

    #include <stream.hpp>
    int main() {
        char c;
        while(cin.get(c)) cout.put(c);
        return 0;
    }

does not completely copy stdin to stdout! (I believe the error was in
streambuf::sgetc peeking at the next character to determine whether to
set _eof.)  There are other buffering problems as well.  Sigh!

The compiler also bitches at files that do not explicitly end with a
newline, although ZED occasionally generates such beasts.  Why is this 
a lexical error?

Installation went smoothly, and the compiler is very fast, but I'll
reserve judgement until I have all the work arounds to get my code
working again.
-----
distressed,
jeffa