Hi Fraser, I tested this on my linux box and things seemed to work there. Not sure whether there are any unpleasant surprises on the WIndows system or whether the below example is special in any sense and I simply didn't run across this when doing my testing. But since you say that the arrows are merely pointing in the wrong direction this shouldn't be to hard to fix/figure out. I am traveling right now and probably won't find the time to fix this properly before the release, so for now I will revert the default arrow shape back to the old '"vee" type and push a fixed and more thoroughly tested version back to the release once the smoke has settled. Thanks for the notice, Florian On 18.04.2009, at 16:14, Sim, Fraser wrote: > Hi Florian, > > The new normal arrows look great but their orientation is not correct > for some of the egdes. I ran the same code as below in 2.9.0. The > edges > a~d, d~e, b~d are fine but a~b, c~d and c~e are reversed with the > arrow > pointing in the wrong direction back up the edge. > > I took a look at the source code but couldn't see the problem > immediately but didn't know what the 'out' variable meant and why it > was > necessary for the normal arrow but none of the other arrowhead > types. It > looked like if it was set the arrow would rotate 180 degrees and maybe > cause the problem but I'm just guessing here. > > The rankdir problem is fixed, thanks. > > Thanks for your help, > Fraser > > Note: I am using graphviz version 2.21.20090302.545 because 2.22 > didn't > work with your code. Earlier versions of graphviz crash on Vista 64- > bit. > The Rgraphviz binary did not work on my system so I had to build the > new > version (1.21.11) from source. It built fine with no errors and I can > make graphs ok. > > Here's my sessionInfo: > R version 2.9.0 (2009-04-17) > i386-pc-mingw32 > > locale: > LC_COLLATE=English_United States.1252;LC_CTYPE=English_United > States.1252;LC_MONETARY=English_United > States.1252;LC_NUMERIC=C;LC_TIME=English_United States.1252 > > attached base packages: > [1] grid stats graphics grDevices datasets utils > methods > > [8] base > > other attached packages: > [1] Rgraphviz_1.21.11 graph_1.21.7 rcom_2.1-3 > rscproxy_1.3-1 > > > loaded via a namespace (and not attached): > [1] cluster_1.11.13 tools_2.9.0 > > -----Original Message----- > From: Florian Hahne [mailto:fhahne at fhcrc.org] > Sent: Thursday, April 16, 2009 7:31 PM > To: Sim, Fraser > Cc: bioconductor at stat.math.ethz.ch > Subject: Re: [BioC] RGraphviz > > Sim, Fraser wrote: >> Hi Florian, >> >> Here is some example code: >> >> library(Rgraphviz) >> set.seed(123) >> V <- letters[1:5] >> M <- 1:2 >> g1 <- randomGraph(V, M, 0.5) >> edgemode(g1) <- "directed" >> x <- layoutGraph(g1) >> >> edgeRenderInfo(x) = list(arrowhead = "tee") >> renderGraph(x) >> >> edgeRenderInfo(x) = list(arrowhead = "normal") >> renderGraph(x) >> >> The first graph rendering is correct with arrowhead converted to > tee's. >> The second generates open arrowheads more like the 'vee' according to >> the graphviz documentation. Also if you run the code with 'vee' there > is >> no difference from 'normal'. >> > The base R graphics arrow function doesn't know how to draw filled > arrow > > heads. And since all Rgraphviz plotting is done using base graphics, > the > > "normal" arrow shape was more like the 'vee' shape in Graphviz. I > added > filled arrowheads to the latest Rgraphviz devel version (1.21.11), and > these are used as the default now. Things get a bit ugly once you > start > resizing the X11 device, but that problem was there before, it's just > easier to spot for the filled arrow heads. Note that you will only get > those when you use the new layoutGraph/renderGraph interface. >> It appears that some of the other shapes are also not supported and >> revert back to the standard. I understand that not all shapes may be >> supported but it would be more intuitive if the renderGraph produced > an >> error or warning for non-supported shapes rather than just silently >> reverting them to an open arrow. >> > You will now get a warning when using a non-supported arrow head type. > We still fall back to the default arrowhead, which is a good thing, I > think. > > Bw, > Florian >> Thanks, >> Fraser >> >> -----Original Message----- >> From: Florian Hahne [mailto:fhahne at fhcrc.org] >> Sent: Tuesday, April 14, 2009 7:09 PM >> To: Sim, Fraser >> Cc: bioconductor at stat.math.ethz.ch >> Subject: Re: [BioC] RGraphviz >> >> Hi Fraser, >> could you provide a reproducible example for us to take look. From >> your post I can't tell how or where you tried to set the symbols. >> Cheers, >> Florian >> >> >> On 14.04.2009, at 14:09, Sim, Fraser wrote: >> >> >>> Hi all, >>> >>> I am trying to set custom arrowheads in a graphviz-based plot. It I >>> set >>> to "tee" or "dot", I get the expected symbols. >>> >>> However, if I use "normal", I do not get a closed arrowhead and >>> there is >>> no difference between "normal" and "vee" arrowheads. >>> >>> Any idea why? >>> >>> Cheers, >>> Fraser >>> >>> sessionInfo: >>> R version 2.8.1 (2008-12-22) >>> i386-pc-mingw32 >>> >>> locale: >>> LC_COLLATE=English_United States.1252;LC_CTYPE=English_United >>> States.1252;LC_MONETARY=English_United >>> States.1252;LC_NUMERIC=C;LC_TIME=English_United States.1252 >>> >>> attached base packages: >>> [1] grid stats graphics grDevices utils datasets >>> methods >>> >>> [8] base >>> >>> other attached packages: >>> [1] Rgraphviz_1.21.7 graph_1.20.0 rcom_2.0-4 > rscproxy_1.0-12 >>> >>> loaded via a namespace (and not attached): >>> [1] cluster_1.11.12 tools_2.8.1 >>> >>> _______________________________________________ >>> Bioconductor mailing list >>> Bioconductor at stat.math.ethz.ch >>> https://stat.ethz.ch/mailman/listinfo/bioconductor >>> Search the archives: >>> >> http://news.gmane.org/gmane.science.biology.informatics.conductor >> >> > > > -- > Florian Hahne, PhD > Computational Biology Program > Division of Public Health Sciences > Fred Hutchinson Cancer Research Center > 1100 Fairview Ave. N, M1-B514 > PO Box 19024 > Seattle, Washington 98109-1024 > 206-667-3148 > fhahne at fhcrc.org >