Re: [boost] 1.16.1 Posted

From: Lois Goldthwaite (loisg_at_[hidden])
Date: 2000-07-06 10:40:52

Would it be worth mentioning the specific non-compliant feature in the error
message?  -- ".... because xxx is required in order to achieve yyy ..."  I know
you can dig this out of the code, but it doesn't look like any more trouble if
you're writing a specific error message anyway, and it might help the
documentation effort. Or guide the user of the library on how to work around
the limitation. Some of us have not yet achieved guru-hood on the thornier bits
of C++, especially if we can't practice with them because of broken
compilers.   :-)

Lois

Beman Dawes wrote:

>  >Note that another compiler and several more test programs have been added
>
>  >to
>  >http://www.boost.org/libs/compiler_status.htm.  More of the compiles are
>  >now succeeding, thanks to the fixes in this release from library
>  >maintainers.
>
> In private email John Maddock mentioned that call_traits_test.cpp isn't
> supposed to compile with compilers (VC++ in particular) which don't support
> partial specialization.
>
> Fair enough; we have been saying all along that boost doesn't require
> libraries to support broken compilers, particularly if a workaround is
> costly or impossible.
>
> What would be helpful to users IMO is if such libraries reported the
> problem rather than just letting the compiler produce misleading error
> messages.  For example,
>
> # ifdef BOOST_NO_TEMPLATE_PARTIAL_SPECIALIZATION
> #   error Library cannot be used with this non-standard-compliant compiler
> # endif
>
> If others think this is worthwhile, I will add it to the guidelines.
>
> --Beman
>
>
> ------------------------------------------------------------------------
> Where do sports heroes like Derek Jeter, Mia Hamm,
> Vince Carter and Peyton Manning hang out? Where else?
> Click now and find ‘em all here!
> http://click.egroups.com/1/6211/4/_/9351/_/962888292/
> ------------------------------------------------------------------------