Re: [Spirit Development] Bug using operator >> and > in same sequence

-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1

Hey - I asked heller about this the other day, here's his reply:

Begin forwarded message:

Date: Sun, 2 Jan 2011 10:53:59 +0100
From: Thomas Heller <[email protected]>
To: [email protected]
Subject: spirit + flatten tree + SAs


Hi Bryce,

Sorry about yesterday ... needed to go real quick.
Anyway, I looked into the issue. You are right, proto::flatten only flattens 
expressions with the same tag type, in a way, here is the excerpt from the 
docs:
        /// \brief A PolymorphicFunctionObject type that returns a 
"flattened"
        /// view of a Proto expression tree.
        ///
        /// A PolymorphicFunctionObject type that returns a "flattened"
        /// view of a Proto expression tree. For a tree with a top-most node
        /// tag of type \c T, the elements of the flattened sequence are
        /// determined by recursing into each child node with the same
        /// tag type and returning those nodes of different type. So for
        /// instance, the Proto expression tree corresponding to the
        /// expression <tt>a | b | c</tt> has a flattened view with elements
        /// [a, b, c], even though the tree is grouped as
        /// <tt>((a | b) | c)</tt>.

However, i do not believe that this is a bug in spirit. It is more of a 
pitfall. Spirits automagic attribute calculation stuff seems to synthesize 
the attribute just fine. That means that this works as expected:

  rule<iterator, vector3<int, int, int>(void)>
    r = (int_ >> int_ > int_);

Let me remind you that semantic actions disable this automatic attribute 
stuff. That means, that the above rule sees a different attribute (this is a 
little counter intuitive, but on a closer look just right).

  rule<iterator, vector3<int, int, int>(void)> 
    r = (int_ >> int_ > int_) // "raw" synthesized attribute is of type:   
    // boost::fusion::vector2<boost::fusion::vector2<int, int>, int>
        [
            _val
               = construct<vector3<int, int, int> >
                 ( // manually decompose it:
                   phoenix::at_c<0>(_1)
                 , phoenix::at_c<1>(_1)
                 , _2
                 )
        ];

Hope that helps,
Thomas


- -- 
Bryce Lelbach aka wash
boost-spirit.com

-----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.9 (GNU/Linux)

iEYEARECAAYFAk0g6BoACgkQO/fqqIuE2t6NhACg13A2GPl2RRNFs1CqSPZzn0Pl
YIcAoND7jnpuQTr0dwmuOVaeYivGsp7e
=7Mf7
-----END PGP SIGNATURE-----
------------------------------------------------------------------------------
Learn how Oracle Real Application Clusters (RAC) One Node allows customers
to consolidate database storage, standardize their database environment, and, 
should the need arise, upgrade to a full multi-node Oracle RAC database 
without downtime or disruption
http://p.sf.net/sfu/oracle-sfdevnl