Re: [boost] [build] request for modification.

From: Ronald Garcia (garcia_at_[hidden])
Date: 2003-02-05 19:26:09

On Wed, 5 Feb 2003, Dave Abrahams wrote:

> On Wednesday, February 05, 2003 5:58 PM [GMT+1=CET],
> Ronald Garcia <garcia_at_[hidden]> wrote:
>
> > On Wed, 5 Feb 2003, Rene Rivera wrote:
> >
> > > [2003-02-05] Ronald Garcia wrote:
> > > >
> > > > I would like to request the addition of the -ansi flag to the
> > > > [ ... ]
>
> It is our usual policy to run in "most conforming, plus extensions mode".
> IOW, extensions like long long and platform-specific capabilities like
> __cdecl are typically enabled when they're available.
>
> What does -ansi add/remove from the capabilities of Intel C++?
>

I'm afraid I don't know all of the details, but I was specifically
interested in it's noticing the use of values from template-dependent base
classes.  A little while ago, there was some discussion of
bugs in multi_array related to this issue.  I discovered then that of the
compilers I currently use, icc with the "-ansi" flag was the only one that
detected this error.  I (hopefully!) fixed those bugs, which are in a
sense resulting from compiler extensions (i.e. the EDG front-end can be
smart enough to dive into dependent base classes to find info).

I'm not sure what your thoughts on this "feature" wrt Boost.Build is.
Perhaps Boost.Build could use a flag marking certain levels of strictness.
Were that the case, the Boost regression test system could enforce a
higher level of strictness than vanilla Boost.Build....or we could leave
it as is and I could just ignore future bug reports ;-).

Cheers,

ron