[boost] Re: config.h[pp]

From: Andy Glew (glew_at_[hidden])
Date: 1999-07-14 13:04:48

> Let me put it this way: I compile all my code with a real existing,
> publically available compiler (egcs-1.1.2), use only standard code
> (somewhat ascertained by '-pedantic -ansi -W -Wall') and no compile
> time switching as long as I'm working on stuff which is supposed to
> work with a standard conforming compiler

Interesting.

I use EGCS as an example of a not really very standards conforming compiler.

It's getting better quickly, but I keep running into problems with unimplemented
features of the standard when friends from Intel send me code to run.
They tend to run Intel's C++ compiler, and occasionally MS VC++.
I use EGCS 1.1.2 today, and usually keep up to the latest.

Now, let's see, what areas has EGCS been deficient in?

+ namespaces - hah! it's fixed now!

        For a long time GNU compilers did not support multiple namespaces,
        resulting in a lot of kluges to get around this basic need.  I am very happy
        to now know that this is fixed. (Just ran a simple test program.)

+ member offsets as template arguments

        Unfortunately, I don't have the test program for this. If it is fixed, I will
        be tremendously happy, since it is the last obstacle to some truly generic code.

Well, I hope that all the deficiencies have been corrected. Is there a list somewhere
of where EGCS stands wrt implementing the full ANSI C++ standard?

------------------------------------------------------------------------

eGroups.com home: http://www.egroups.com/group/boost
http://www.egroups.com - Simplifying group communications