"Yannick Duchêne (Hibou57)" wrote in message news:op.w0hmrfoqule2fv@cardamome... Le Tue, 16 Jul 2013 02:35:16 +0200, Randy Brukardt a écrit: > "Dmitry A. Kazakov" wrote in message > news:pgggo3irufxz$.mbvohwd9o6qd.dlg@40tude.net... >> On Sun, 14 Jul 2013 23:36:22 -0700 (PDT), Peter Brooks wrote: >> >>>> Alternatively, is there a good Ada package for handling directed >>>> graphs? >>> >>> Well, it is difficult to provide a universal implementation of graphs. >>> Usually, you would needs a custom implementation. >> >> Right. If your graph doesn't have any cycles (that is, it's a tree),[.] >I believe this was an oops: an acyclic graph is not necessarily a tree. I guess you're right. I was thinking that a "join" would have to necessarily cause a cycle, but that isn't true if both nodes are leafs and there are probably more cases as well. What I *was* thinking is that the multi-way tree that is contained in Ada.Containers supports a lot more structures than just the classic binary tree, and that many (maybe most?) acyclic graphs that you'll encounter in practice can be mapped into it. (I've never needed a more complex graph than a multi-way tree in any of my programs. YMMV.) Randy.