RE: [boost] Re: Using ggcl with Sun CC-6.0 (Forte)

From: Beman Dawes (bdawes_at_[hidden])
Date: 2001-02-02 09:57:18

At 08:29 AM 2/2/2001 -0500, Mishkin, Nat wrote:

 >Ah, I guess I should have made clear that despite my slashing and 
burning,
 >I still got compilation errors.  The Solaris compiler report that Beman
 >pointed me at in another reply to my post also indicates problems with 
this
 >compiler, although that report looks like it's saying it compiled but
 >didn't _run_ correctly.

A failure on a "run" test might be a compile-and-link failure, or an 
execution failure.  You have to look at the logs (look at the CVS status 
page for links now, but they will be visible after the next release). It 
might be nice if the table failure was changed to be more specific, but 
Jens has done so much good work on regression.cpp lately I don't want to 
bother him asking for yet another enhancement.

 >  The number of failures shown in the report for _all_ the
 >different compilers on Solaris makes for pretty depressing reading.

Boost is just starting to include Solaris in testing.  What usually happens 
over time is that boost developers add workarounds that increase the 
success rate.

 >It looks like I'll need to considerably limit my ambitions as far as use 
of
 >modern C++ capabilities if I want to have a portable application.  (Sigh.
 >Didn't this language get standardized a fairly long time ago?  I guess 
all
 >the compiler folks are abandoning the antique languages like C++ and are
 >off working on Java or C# :-)

Maybe the release of gcc 3.0.0 (due this quarter) will put pressure on some 
of the commercial vendors.  At least I hope so.

--Beman