R devel - May 2013 - Bug in stats::plot.hclust/graphics:::plotHclust/underlying C code (C_dend) in R-3.0?

Hi all,

it appears that plot.hclust disregards the argument 'cex'. Up until
2.15.1 (and probably all 2.15.x), specifying cex in plot.hclust
changed the size of the plotted labels on each leaf. In 3.0.0 the
character size remains the same irrespective of the cex setting. Since
the help text for plot.hclust does not mention the cex argument, it is
possible that an "undocumentde feature" was simply removed - but it
was very useful when it was available.

The following code illustrates the problem:

data = c(1:20)
dst = dist(data)
tree = hclust(dst, method = "a")
plot(tree, cex = 1)
plot(tree, cex = 0.5)
plot(tree, cex = 0.2)

Under R-2.15 and earlier, the leaf labels would get progressively
smaller; under 3.0.0, the label size remains the same. I tried this
both under Windows

R version 2.15.1 (2012-06-22)
Platform: x86_64-pc-mingw32/x64 (64-bit)

locale:
[1] LC_COLLATE=English_United States.1252
[2] LC_CTYPE=English_United States.1252
[3] LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C
[5] LC_TIME=English_United States.1252

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base

-------------------------- Session info for R-3.0.0

R version 3.0.0 (2013-04-03)
Platform: x86_64-w64-mingw32/x64 (64-bit)

locale:
[1] LC_COLLATE=English_United States.1252
[2] LC_CTYPE=English_United States.1252
[3] LC_MONETARY=English_United States.1252
[4] LC_NUMERIC=C
[5] LC_TIME=English_United States.1252

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base

and under Linux

R version 3.0.0 Patched (2013-04-24 r62658)
Platform: x86_64-unknown-linux-gnu (64-bit)

locale:
 [1] LC_CTYPE=en_US.UTF-8       LC_NUMERIC=C
 [3] LC_TIME=en_US.UTF-8        LC_COLLATE=en_US.UTF-8
 [5] LC_MONETARY=en_US.UTF-8    LC_MESSAGES=en_US.UTF-8
 [7] LC_PAPER=C                 LC_NAME=C
 [9] LC_ADDRESS=C               LC_TELEPHONE=C
[11] LC_MEASUREMENT=en_US.UTF-8 LC_IDENTIFICATION=C

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base

Looking at the source code of plot.hclust leads to
graphics:::plotHclust which calls .External.graphics, which is where
my ability to read code quickly stops :)

I realize that these are not the most recent releases but I couldn't
find anything mentioning a bug fix for this in the changelog, so I
assume it has not been fixed.

Thanks,

Peter

>>>>> Peter Langfelder <peter.langfelder at gmail.com>
>>>>>     on Fri, 31 May 2013 16:32:06 -0700 writes:
    > Hi all, it appears that plot.hclust disregards the
    > argument 'cex'. Up until 2.15.1 (and probably all 2.15.x),
    > specifying cex in plot.hclust changed the size of the
    > plotted labels on each leaf. In 3.0.0 the character size
    > remains the same irrespective of the cex setting. Since
    > the help text for plot.hclust does not mention the cex
    > argument, it is possible that an "undocumentde feature"
    > was simply removed - but it was very useful when it was
    > available.

    > The following code illustrates the problem:
>    data = c(1:20)
>    dst = dist(data)
>    tree = hclust(dst, method = "a")
>    plot(tree, cex = 1)
>    plot(tree, cex = 0.5)
>    plot(tree, cex = 0.2)

    > Under R-2.15 and earlier, the leaf labels would get
    > progressively smaller; under 3.0.0, the label size remains
    > the same. I tried this both under Windows

    > R version 2.15.1 (2012-06-22) 

 [............]
    

    > R version 3.0.0 (2013-04-03)

 [............]


    > and under Linux

    > R version 3.0.0 Patched

[...........]

    > Looking at the source code of plot.hclust leads to
    > graphics:::plotHclust which calls .External.graphics,
    > which is where my ability to read code quickly stops :)

    > I realize that these are not the most recent releases but
    > I couldn't find anything mentioning a bug fix for this in
    > the changelog, so I assume it has not been fixed.

You are right.    Duncan Murdoch found the place (in the C code
indeed), and I've found and fixed the small inconsistency that
happened (between 2.15.x  and 3.0.0) when that code was slightly
streamlined and moved to the 'graphics' package.

    > Thanks,
    > Peter

Thank you Peter,  for reporting it in such a precise manner!

Martin Maechler, ETH Zurich