@OleksandrR. Thanks a lot for highlighting the PDF method in that post. I was able to use text characters from Insert | Special Character... for the checkmark and the cross.

@Szabolcs oh dear, are Mathematica's graph functions so bad you decided on a full reimplementation?

Since you asked me about licences before... I noticed that this IGraph library uses the GPL for the software and the GFDL for its documentation. This is the same arrangement as for emacs. But the hilarious thing is that these two are totally incompatible with each other! So the manual cannot quote any of the source code, and no code examples given in the manual can be used with the software (if you want to distribute it). It is a total nightmare.

I doubt this will affect you in any way, though.

@OleksandrR. I really start to wonder what you do at night!? Are you reading license specs for fun?

@OleksandrR. I tried the BoundaryDiscretizeGraphics method with the PolygonMarkers package but it does not produce the correct polygon.

@Edmund it must be a closed, non-self-intersecting polygon. No holes.

@Edmund I don't have version 10 installed at the moment so I can't test it. I would use the PDF method if I were you

@OleksandrR. Okay. Thanks.

@halirutan to tell you the truth, I just find it amusing when people choose licences as a way to express themselves politically and then end up tying themselves up in knots with incompatible choices.

@OleksandrR. I think it has to do with the order of the points that you get in with this method. The points don't "follow" a line around the text. Instead they jump around the edges of the text. This produces a wild graphic when PolygonMarker plots it.

@Mr.Wizard I cast my delete vote on it. The first one can't be voted on until the second one is deleted.

@halirutan I really tried hard with WU: via bank transfer, using two different credit cards... they simply did not want my money.

(uh, non sequitur... wrong room, early in the morning)

recursively until no replace are left?

@William Either use ReplaceRepeated (//.) or apply FixedPoint to your expression.

thank you!

@OleksandrR. @Leonid I shouldn't have ranted that much the other day about Graph. To be fair, Mathematica's collection of graph routines is pretty good. There aren't that many packages out there that pull together so much functionality. What's frustrating is that bugs don't seem to get fixed unless people complain about them, that there are things that look fairly fundamental (especially related to modification of graph objects, the basic Property API) and ...

... are still bugged 4 releases after Graph was introduced, that there seem to be some stubbornness about fixing usability problems like the NeighborhoodGraph thing, and the thing with how multigraphs seem to have been introduced recklessly.

@Oleksandr But igraph is not without bugs either, I discovered several recently. What I was doing with igraph is definitely not a replacement for Mathematica's graph stuff. It's for complementing it. I'm mostly looking at things that are not available in Mathematica and a few things that are hard to verify without a second implementation.

Graph functions are not unique in that they sometimes have bugs and return wrong results, but they're a bit different from some other areas in that the results are hard to verify. With e.g. eigenvectors verification is as simple as checking if vec / A.vec is a constant vector. Computation is hard but verification is easy. With graph stuff very often verification is hard too.

It might look like some of the things that I included so far in that igraph package are already present in Mathematica, but there's more than meets the eye. Graph functions have to support undirected & directed graphs, edge-weighted and unweighted graphs, graphs with self loops, multiple edges, maybe even mixed directed and undirected edges when that might make sense ... Typically functions only support a subset of these. igraph and Mathematica might not implement the same subset.

That's why I was interested in igraph. For example, Mathematica was weighed edge betweenness, but it only has unweighted vertex betweenness.

Also, again to be fair, Mathematica does tend to be faster than igraph on many tasks.

Many of Mathematica's problems concern modification of graph objects and attributes (e.g. weights). I didn't touch on these at all in IGraph/M. I'm not exposing the igraph data type to the user (though there's an internal interface to it). Instead public functions just convert Graph to an igraph format, then convert the result back. Thus it makes no sense to implement stuff like attributes, edge and vertex deletion, etc. as this would end up being equivalent to converting to an edge-list ...

manipulating that and converting back. igraph wouldn't even get involved.

@Mr.Wizard the second one was deleted now so I cast my vote on the first as well

@YvesKlett It seems we live in different time zones. I was just arrived in bed when you wrote :-)

Might this be solvable using the (new in v10) GraphAutomorphismGroup? First find out if they're isomorphic without colouring, then check if the automorphism group has any elements that leave the colours in place.

Anyone experienced with graph theory?

Hi

@barznjy Have you looked in the Edit menu? There's Redo there.

@OleksandrR. Thanks for your help!

Has anyone had any problems with BeginPackage not loading a sub-sub-package

I have a setup where Package1 calls BeginPackage with two items in the needs list; Package11 and Package12. Each of these have one package in thier needs list; Package111 and Package121 respectively.

When I call Get (<<) on Package1 it loads without error. However, when I call a function in Package1 that is from Package111 it gives an error that it can't find it

Why is this the case since Package1 loads Package11 which loads Package111

@Szabolcs there is no Redo

@Szabolcs please look at the above list

@barznjy it should be there. Does Ctrl+Y work for it? If so then the menu item can be reinstated if necessary

If I were you, I'd ask about this on the site

I mean, re-adding the Redo menu item. Verify that redo works first

@SjoerdC.deVries good point. Prior to version 10 there was no redo menu item.

@OleksandrR. Right. IIRC redo arrived together with multiple undo.