Re: [boost] Patches for unversioned MSVC checks

From: David Abrahams (david.abrahams_at_[hidden])
Date: 2002-07-13 12:38:46

In fairness, though, only a very few of the uses of BOOST_MSVC that we have
actually fall into that category.

----- Original Message -----
From: "Jeremy Siek" <jsiek_at_[hidden]>

> Hi Phil,
>
> That is indeed the better approach. However, you have to remember that
> sometimes we are not dealing with a missing feature, but a feature that
> breaks down in strange and complex ways. Without looking at the source
> code for the compiler, and finding the bug, it is hard to classify the
> problem and come up with a macro for it. In these situations we often
fall
> back on just using the macro for the compiler.
>
> Cheers,
> Jeremy
>
> On Sat, 13 Jul 2002, Phil Nash wrote:
> phil.n> Shouldn't these things go in the config files as macros for each
compiler
> phil.n> dependant feature, rather than compiler conditionals being
exposed in the
> phil.n> program code?
> phil.n> This way the features in question only need to be updated in one
place. At
> phil.n> least that's my understanding of it. I'm not saying *you* should
have to do
> phil.n> this, David, I'm just wondering why we have this problem in the
first place
> phil.n> since we seem to have a pretty good way of dealing with exactly
this
> phil.n> complexity already.
> phil.n>
> phil.n> Regards,
> phil.n>
> phil.n> [)o
> phil.n> IhIL..
> phil.n>
> phil.n> ----- Original Message -----
> phil.n> From: "David Abrahams" <david.abrahams_at_[hidden]>
> phil.n> To: "boost" <boost_at_[hidden]>
> phil.n> Sent: Saturday, July 13, 2002 1:37 PM
> phil.n> Subject: [boost] Patches for unversioned MSVC checks
> phil.n>
> phil.n>
> phil.n> >
> phil.n>
> phil.n>
>
phil.n> -------------------------------------------------------------------
---------
> phil.n> ----
> phil.n>
> phil.n>
> phil.n> > I've taken the liberty of committing fixes for *some* of
boost's
> phil.n> BOOST_MSVC
> phil.n> > checks which weren't checking the version number. I took a
somewhat
> phil.n> > conservative route, which also means it's somewhat
ham-fisted -- in many
> phil.n> > cases the original test was written with MSVC6 in mind, and I
just assumed
> phil.n> > that the same workaround would be needed for MSVC7 but not for
later
> phil.n> > compilers. Hopefully, the enclosed list will be enough to
prompt authors
> phil.n> > and maintainers to look over the changes and tune them
appropriately.
> phil.n> >
> phil.n> > Note: in case it wasn't clear:  I DID NOT TRY TO FIX
EVERYTHING. In
> phil.n> > particular, I gave up quickly on the graph library examples
directory
> phil.n> > because it was too big and I didn't touch regex at all for lack
of
> phil.n> > expertise. For other libraries I only made a change where the
answer
> phil.n> seemed
> phil.n> > obvious to me. I double-checked my work, and I hope I didn't
break
> phil.n> > anything. No guarantees of course.
> phil.n> >
> phil.n> > -Dave
> phil.n> >
> phil.n> > P.S. I am certain that we have lots of similar issues with
__BORLANDC__
> phil.n> and
> phil.n> > __MWERKS__, probably others as well.
> phil.n> >
> phil.n> >
> phil.n> >
+---------------------------------------------------------------+
> phil.n> >                   David Abrahams
> phil.n> >       C++ Booster (http://www.boost.org)               O__  ==
> phil.n> >       Pythonista (http://www.python.org)              c/ /'_ ==
> phil.n> >   resume: http://users.rcn.com/abrahams/resume.html  (*) \(*)
==
> phil.n> >           email: david.abrahams_at_[hidden]
> phil.n> >
+---------------------------------------------------------------+
> phil.n> >
> phil.n>
> phil.n>
>
phil.n> -------------------------------------------------------------------
---------
> phil.n> ----
> phil.n>
> phil.n>
> phil.n> > _______________________________________________
> phil.n> > Unsubscribe & other changes:
> phil.n> http://lists.boost.org/mailman/listinfo.cgi/boost
> phil.n> >
> phil.n>
> phil.n> _______________________________________________
> phil.n> Unsubscribe & other changes:
http://lists.boost.org/mailman/listinfo.cgi/boost
> phil.n>
>
> ----------------------------------------------------------------------
>  Jeremy Siek                          http://php.indiana.edu/~jsiek/
>  Ph.D. Student, Indiana Univ. B'ton   email: jsiek_at_[hidden]
>  C++ Booster (http://www.boost.org)   office phone: [phone]
> ----------------------------------------------------------------------
>
>
> _______________________________________________
> Unsubscribe & other changes:
http://lists.boost.org/mailman/listinfo.cgi/boost