Re: [Boost-users] Fusion transform question

From: Christian Holmquist (c.holmquist_at_[hidden])
Date: 2008-07-20 14:47:28

2008/7/19 Steven Watanabe <watanabesj_at_[hidden]>:
> AMDG
>
> Joel FALCOU wrote:
>>
>> I just read the parts about the lazyness of fusion transformation. It
>> indeed escpaed me for a few seconds.
>> So basically the fusion transform are more suited for the runtime aprt of
>> the code rather than the compile-time one ?
>
> Right, fusion::transform is appropriate for mixed compile-time/run-time
> calculations.
> MPL is more appropriate for purely compile time constructs.

Maybe this could be stated more clearly in the docs? The last part of
fusion introduction section says:
 "For example, there are times when it is convenient to work solely on
MPL  using pure MPL  sequences, then, convert them to Fusion sequences
as a final step before actual instantiation of real runtime objects
with data"

This little line got stuck into my head, and at least in my experience
using mpl algorithms&sequences is not only more convenient but also
required most of the time, prior to instantiate any fusion sequence to
do the runtime part.

Adding something like:
!
'Choose MPL over fusion when doing pure type calculations. Once the
static type calculation is finished, you can instantiate a fusion
sequence (link to result_of::as_vector<S>') for the runtime part.'

My 2 c..,

/ Christian Holmquist

>
> In Christ,
> Steven Watanabe
>
> _______________________________________________
> Boost-users mailing list
> Boost-users_at_[hidden]
> http://lists.boost.org/mailman/listinfo.cgi/boost-users
>