[boost] Re: int_const_guidelines out of date?

From: Markus Werle (numerical.simulation_at_[hidden])
Date: 2003-11-26 09:34:40

David Abrahams wrote:

> Markus Werle <numerical.simulation_at_[hidden]> writes:
> 
>> OK, I give a chance to everyone:
>>
>> For every compiler: if 30 users answer this post
>> and declare that they are really interested in
>> Daixtrose to be boostified and that they have no chance
>> to switch  compilers within the next 2 years, then I will
>> try to support that one.
>>
>> Otherwise it's Intel-C++-7.1/8.0 and gcc-3.4.
> 
> Suit yourself.  Be forewarned that the compilers supported may affect
> the community's interest level.  If you don't care about having lots
> of users, you can certainly get by with that strategy.

I see.
Is it much better if MSVC-7.x (x>1) is added to the set
of possible compilers?

Which minimal set of the macros mentioned 
in the int_const_guidelines do you think
is mandatory then?

Probably I should take the same approach as loki,
where there is a pure version and a portable version
side by side, dispatched by 
BOOST_ETL_USE_PORTABLE_VERSION?

> BTW, we have lots of libraries you can use to eliminate portability
> hacks in your own code.  For example, I have way fewer platform
> headaches now that I use the MPL for all my metaprogramming jobs. 

I really want etl be based on boost::mpl as far as possible.
(though I sometimes suffer under the brevity of the docs,
a few more examples would help).

Is there a detailed documentation of all the mpl macros?

> I daresay that between MPL, type traits, and possibly Fusion 

Fusion? Is that another library? Or do you mean fusion
of mpl and typetraits?

> you could
> make writing an expression template library seem easy.

Let's say  "cleaner in design". Easy assumes I already
have learned all that is available.

Markus

P.S.: do you see any really bad design flaws in Daixtrose
which should be avoided in etl?

-- 
Compile time analytic differentiation with Daixtrose!
Visit http://daixtrose.sourceforge.net/