From: Jeremy Siek (jsiek_at_[hidden])
Date: 2004-03-22 17:58:48
On Mar 16, 2004, at 3:47 AM, HIRAKI Hideaki wrote:
> Hello there,
> I submitted a bug fix on read_graphviz() to the bug tracker (Bug
> ID 865882, see P.P.S. below for the summary). When thinking about
> possible fixes, I got some questions.
> Q1) May boost depend on build tools other than C++ compiler?
> There seemed to be a building problem due to the dependency of the
> parser code on some specific versions of bison. This is not a real
> problem because the generated C++ codes are bundled. I think the
> dependency should be removed, if possible.
That sounds reasonable.
> Q2) How easy is it to migrate a flex/bison code to spirit?
> Boost seems to have its own parser generator, spirit. Its features
> are uncertain for me. I just expect that it is straightforward for
> an expert of spirit to write a reentrant parser to be replaced with
> the current flex/bison parser.
I am also not an expert with the spirit parser, but that sounds like a
direction. Another approach would be to just write the parser by hand
> Q3) Why boost::subgraph isn't general subgraph but "induced" subgraph?
Hmm, I thought graphviz used induced subgraphs. I'll look into this.
> Q4) How are the graph adapters for graphs of graphviz library?
> Another option imagined is wrapping the graphviz library. This will
> introduce library dependency instead of build tool dependency. But no
> parser need to be developed. A data structure for general subgraph
> will be provided as a wrapped object. BGL seems to include wrappers for
> Stanford GraphBase and LEDA, C++ libraries. How are they adapted to
> graphviz, a C library?
That's an interesting idea. I bet it would be straightforward to
such an adaptor, so if you were to implement it, it would be a welcome
addition to the BGL.
Boost list run by bdawes at acm.org, gregod at cs.rpi.edu, cpdaniel at pacbell.net, john at johnmaddock.co.uk