Re: [BGL] build problems

Angus Leeming writes:

David Abrahams wrote:

...

Angus Leeming writes:

...

Dear all,

I'm a bjam newbie trying to build the BGL in boost 1.30.0 on a DEC Alpha running linux. However, the build process is failing to generate the graphviz parser and lexer:

$ cd libs/graph/build $ bjam "-sTOOLS=gcc" ...found 437 targets... ...using 2 temp targets... ...updating 3 targets... ...using graphviz_graph_lex.o... ...using graphviz_digraph_lex.o... gcc-C++-action ../../../libs/graph/build/bin/liblibbgl-viz.a/gcc/debug/inlining-on/runtime-link-dynamic/graphviz_graph_parser.o graphviz_parser.y:44:2: #error Need to define the GRAPHVIZ_GRAPH

Try bjam -sTOOLS=gcc "-sBUILD=<define>GRAPHVIZ_GRAPH=GraphvizGraph"

HTH,

Thank you, Dave. It does indeed fix that particular problem. Now I have another that appears similar:

gcc-C++-action ../../../libs/graph/build/bin/liblibbgl-viz.a/gcc/debug/runtime-link-dynamic/graphviz_graph_parser.o graphviz_parser.y:44:2: #error Need to define the GRAPHVIZ_DIRECTED macro to either 0 or 1

Needless to say, neither "-sBUILD=<define>GRAPHVIZ_GRAPH=GraphvizGraph <define>GRAPHVIZ_DIRECTED=GraphvizDirected" nor "-sBUILD=<define>GRAPHVIZ_DIRECTED=GraphvizDirected" (in case I got the syntax wrong ;-)

have any effect.

Why is that needless to say? Both of those seem to have the intended effect. What do you see on the generated command-lines?

I don't know how familiar you are with the code here, but the Makefile way would be to define specialised targets (my terminology is probably incorrect), in transparent terms something like:

graphviz_graph_parser.o: graphviz_parser.C g++ -DGRAPHVIZ_DIRECTED=0 -o graphviz_graph_parser.o graphviz_parser.C

graphviz_digraph_parser.o: graphviz_parser.C g++ -DGRAPHVIZ_DIRECTED=1 -o graphviz_digraph_parser.o graphviz_parser.C

ie, two object files from one source file. The complication here, of course, is that the source code is in graphviz_parser.y.

That's not complicated, IMO.

Does this enable you to 'cure' the problem or is it fundamental?

I'm sorry, I don't know how to "cure" the problem right now, because I'm not sure what results would be produced by a complete cure. Do you expect to end up with both of these .o files in in the same library? -- Dave Abrahams Boost Consulting www.boost-consulting.com