Re: Regression tests on IRIX and Compaq Tru64 Unix

From: William Kempf (sirwillard_at_[hidden])
Date: 2000-12-13 16:28:14

--- In boost_at_[hidden], rwgk_at_m... wrote:
> P.S.:
> >Downside: it might appear boost was blessing these broken 
> implementations.
> 
> Can someone name an implementation that is not broken according to 
the
> standards implied in this sentence? My most frustrating experience 
as 
> a
> new-comer to C++ is that there does not seem to be such a thing at 
> this
> time.
> The IRIX compiler is based on the EDG frontend. Doesn't this have a
> very good reputation? It seems a bit unfair to disqualify the IRIX
> compiler as a broken implementation only because the <cwhatever> 
files
> are missing.

The EDG front end is used by the compiler to parse the C++ syntax, 
which is totally unrelated to whether or not the compiler provided 
libraries are conformant.  The implementation in question here is the 
implementation of the standard libraries, not of the compiler.

I'm not aware of any compiler that's fully compliant, though most are 
now getting close enough for most uses.  Only older compilers, and 
the upcoming MS compiler, are so far behind in this regard that they 
cause us major problems here.

Bill Kempf