Re: [Boost-users] PropertyGraphConcept concept checking errors for an example implicit graph

Subject: Re: [Boost-users] PropertyGraphConcept concept checking errors for an example implicit graph
From: Jeremiah Willcock (jewillco_at_[hidden])
Date: 2010-06-25 10:02:35

On Thu, 24 Jun 2010, W.P. McNeill wrote:

> Thanks for your help. Â Getting a little closer.
> Once I get the edge properties working I'll look at traversal concepts other than AdjacencyGraphs. Â So far this has been the easy part for me.
> 
> I have been looking at the grid graph, but I couldn't figure out how to add weights to the edges of a grid graph. Â I can't find a simple example of an implicit
> graph with properties. Â (Which is why I'm trying to write one now.)
> 
> What do you mean by "Write an empty struct that inherits from all of the ones that you are going to model"?  I didn't think there were concept base classes to
> inherit from.
> 
> Changing the property typedef in the implicit graph struct to 
> 
> Â Â  Â typedef boost::property<boost::edge_weight_t, float> edge_property_type;
> 
> making the get() functions take constant graph object references, and checking against the ReadablePropertyGraphConcept improves the compilation. Â I now no longer
> have pages of errors, which makes me think I'm on the right track.
> 
> I still have two errors, both of which I'm confused about.
> 
> 1. graph_concepts.hpp:390: error: conversion from ‘EdgeWeightMap’ to non-scalar type ‘boost::typed_identity_property_map<size_t>’ requested
> 
> This is the same as before. Â This error makes me feel like I've got the wrong return value for one of my get() functions, but after double-checking they look
> right.

That does seem odd; could you please send the full error message list 
(including instantiation stacks)?

> 2. property_map.hpp:318: error: no match for ‘operator[]’ in ‘(const boost::typed_identity_property_map<size_t>&)((const
> boost::typed_identity_property_map<size_t>*)(& pa))[k]’
> 
> I didn't think any concept required operator[], so I'm not sure why this is happening.  Line 318 is inside the inline get() function of the put_get_helper(), but
> I'm not using put_get_helper() in my code.

This is probably from the same confusion that triggered the previous 
error; it looks like one of the concept checks is testing against 
MutableLvaluePropertyMapConcept (which does require operator[]), but I 
can't tell why off-hand.

-- Jeremiah Willcock