Re: [boost] [config] BOOST_DEDUCED_TYPENAME

From: Daniel Frey (daniel.frey_at_[hidden])
Date: 2003-03-06 09:05:16

David Abrahams wrote:
> 
> Daniel Frey <daniel.frey_at_[hidden]> writes:
> 
> > I greped for it and it seems it is not used very often. How about using
> > BOOST_WORKAROUND to keep the code local and thus not hide the actual
> > workaround in a MACRO and spread to knowledge? Especially given it's
> > only a workaround for a single compiler. Or do you think it is
> > comparable to BOOST_STATIC_CONSTANT
> 
> I do.

The difference I see is, that BOOST_STATIC_CONSTANT switches between two
legal alternatives, while BOOST_DEDUCED_TYPENAME only hides
non-conforming code (IIUC) which is required for some compilers.

> > and will be used a lot in the future
> 
> Maybe not a lot, but often enough.  I'm sure we have quite a few
> synonyms for it in various headers.  Did you search for just
> "TYPENAME"?

No, I greped for BOOST_DEDUCED_TYPENAME

> I find
> 
>   BOOST_ARG_DEPENDENT_TYPENAME
>   BOOST_MSVC_TYPENAME
>   BOOST_DEDUCED_TYPENAME
>   BOOST_UBLAS_TYPENAME
>   BOOST_SPIRIT_TYPENAME... etc.
> 
> OK, I'm tired of typing these now.
> 
> It really uglifies code to have to add the workaround every place you
> need it.

Agreed. I didn't know about the other MACROs. I just found the one (or
two?) occasions where BOOST_DEDUCED_TYPENAME was used. Generally I think
it's better to have things as local as possible, but if the above
workaround is needed often, it might make sense to keep the macro and if
I understand you correctly, the new macro will replace all of the above
macros, right?

Regards, Daniel

-- 
Daniel Frey
aixigo AG - financial training, research and technology
Schloß-Rahe-Straße 15, 52072 Aachen, Germany
fon: [phone], fax: [phone]
eMail: daniel.frey_at_[hidden], web: http://www.aixigo.de