Re: [Boost-users] Fusion: some concrete questions about maps and vectors

From: Joel de Guzman (joel_at_[hidden])
Date: 2008-08-21 10:44:00

Zeljko Vrba wrote:
> I'm operating on two vectors of pairs of numbers, currently represented by
> std::pair<unsigned, unsigned>.  The first vector is a list of (net,pin) pairs,
> while the second vector is a list of (pin, net) pairs.
> 
> In order to raise a bit the understandability / abstraction level of my code, I
> think that it'd be nice to be able to write at_key<pin>(x) and at_key<net>(x)
> instead of x.first or using tie().
> 
> Looking at fusion manual, it seems that of all the containers, map suits the
> purpose:
> 
> struct pin;
> struct net;
> 
> First question: are Map1 and Map2 different types?
> 
> typedef map<pair<pin, unsigned>, pair<net, unsigned> > Map1;
> typedef map<pair<net, unsigned>, pair<pin, unsigned> > Map2;

Yes.

> According to a simple test program that I wrote, which basically boils down to
> Map m1; Map m2; m1 = m2; they, are different because the compilation fails.
> Hmm, so far, so good. 
> 
> Second question: some algorithms are common both for both types, e.g. sort,
> where I want to have lexicographical comparison between the elements.  Is it
> possible to somehow transform Map1 and Map2 so that they both appear as
> 
> typedef vector<unsigned, unsigned> > Pair

Yes: transform_view. It won't be a fusion vector, but the effect
is what you want.

> with proper lexicographical ordering?  OK, there are conversion metafunctions,
> but I want something more; the following code
> 
>   std::vector<Map1> m;
>   std::sort(m.begin(), m.end());
> 
> should process vector elements (Map1s) _as_if_ they were Pairs or ordinary
> std::pair.  I see there are conversion metafunctions defined in documentation,
> but I don't know how to use them to convert std::vector<Map1> to
> std::vector<Pair> when giving m as argument to sort() [if this is possible
> at all].
> 
> Sorry for the long post and if it is somewhat vague.
> 
> ===
> 
> (I could probably avoid using fusion by manually defining two different struct
> types, each holding pin/net members and defining operator< for each, but I
> want to finally get my hands dirty with fusion on a concrete, not too complex
> problem that I'm facing right now.)

Sounds good. One step at a time. Tell us if you need some help/advice
on various issues.

Regards,
-- 
Joel de Guzman
http://www.boostpro.com
http://spirit.sf.net