R SIG Debian - Jun 2010 - Wrong symbol rendering in plots (Ubuntu)

Hi I am having problems with the rendering of scientific symbols (mu and 
degree) in my plots. Whenever I use these symbols they are rendered 
changed (mu is changed to the proportionality symbol and degree is 
changed to something resembling a gamma) in the X-device. If I make a 
pdf of the plot and open the file in Evince or Okular symbols are also 
rendered wrong, however if I open the file with Xpdf or Acroread they 
are rendered correctly.

I did not have this problem before, it arose after I upgraded both R and 
my system (Ubuntu karmic koala to lucid lynx), so I can not tell for 
sure if the problem is R-related or Ubuntu related (I have posted in a 
Ubuntu forum also http://ubuntuforums.org/showthread.php?t=1325289).

Please let me know if you have any idea of how to fix this, or if you 
can confirm this is not an R-related issue.

Thank you very much in advance,

Eduardo J. Chica
Graduate student
UF/IFAS-CREC

Eduardo J. Chica <ejchica <at> gmail.com>
writes:
> Hi I am having problems with the rendering of scientific symbols (mu and 
> degree) in my plots. Whenever I use these symbols they are rendered 
> changed (mu is changed to the proportionality symbol and degree is 
> changed to something resembling a gamma) in the X-device. If I make a 
> pdf of the plot and open the file in Evince or Okular symbols are also 
> rendered wrong, however if I open the file with Xpdf or Acroread they 
> are rendered correctly.
> 
> I did not have this problem before, it arose after I upgraded both R and 
> my system (Ubuntu karmic koala to lucid lynx), so I can not tell for 
> sure if the problem is R-related or Ubuntu related (I have posted in a 
> Ubuntu forum also http://ubuntuforums.org/showthread.php?t=1325289).
  Could you please post a reproducible example?  It would be
good to see exactly what you are doing, and it will save time
for anyone who wants to try your example on their machine to
try to narrow down the problem.

   Ben Bolker

Hello,

I have the very same problem. Plotting code that used to work before I
upgraded to Ubuntu Lucid Lynx does not work anymore. For example:

plot(1:10)
text(6,4,expression(pi))

The 'pi' greek letter appear as a \neq ("different from"
symbol).

Yvonnick Noel

Hello Ben,
> Does the workaround pointed out later in the thread you're
> responding to (from the last paragraph of a very long 'Note'
> section of ?pdf) help?
>    
Well, I did not try to edit my fonts.conf but I feel this is not a PDF 
issue. I have no problem to have greek letters correctly rendered in 
Ubuntu 10.04 (with LaTeX, or OpenOffice Math for instance with a PDF 
export). This only appears in R plots.

Note that other symbols do not render well too. I put a PDF output of 
demo(plotmath) here:

http://yvonnick.noel.free.fr/wrongsymbolsinubuntu.pdf

Thank you.

Yvonnick

 > sessionInfo()
R version 2.11.0 (2010-04-22)
i486-pc-linux-gnu

locale:
  [1] LC_CTYPE=fr_FR.utf8       LC_NUMERIC=C
  [3] LC_TIME=fr_FR.utf8        LC_COLLATE=fr_FR.utf8
  [5] LC_MONETARY=C             LC_MESSAGES=fr_FR.utf8
  [7] LC_PAPER=fr_FR.utf8       LC_NAME=C
  [9] LC_ADDRESS=C              LC_TELEPHONE=C
[11] LC_MEASUREMENT=fr_FR.utf8 LC_IDENTIFICATION=C

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

[moving to r-devel]

  This PDF looks fine in all of the viewers I have (xpdf 3.02, okular
0.8.2, acroread 8.1.3, evince/poppler 0.8.7) ... I don't really know
enough about PDF/fonts/etc. to debug this ... if I were going to
continue with this I would probably make some kind of minimal plot:
> pdf("test.pdf")
> plot(0:1,0:1,type="n",axes=FALSE,ann=FALSE)
> text(0.5,0.5,expression(alpha %prop% beta))
> dev.off()
(this comes out to only 1397 bytes on my system), make the equivalent
plot with LaTeX

\documentclass{article}
\begin{document}
$$
\alpha \propto \beta
$$
\end{document}

(7514 bytes)

and try to figure out how the fonts are being treated differently ...

Yvonnick Noel wrote:
> Hello Ben,
>> Does the workaround pointed out later in the thread you're
>> responding to (from the last paragraph of a very long 'Note'
>> section of ?pdf) help?
>>    
> Well, I did not try to edit my fonts.conf but I feel this is not a PDF 
> issue. I have no problem to have greek letters correctly rendered in 
> Ubuntu 10.04 (with LaTeX, or OpenOffice Math for instance with a PDF 
> export). This only appears in R plots.
> 
> Note that other symbols do not render well too. I put a PDF output of 
> demo(plotmath) here:
> 
> http://yvonnick.noel.free.fr/wrongsymbolsinubuntu.pdf
> 
> Thank you.
> 
> Yvonnick
> 
>  > sessionInfo()
> R version 2.11.0 (2010-04-22)
> i486-pc-linux-gnu
> 
> locale:
>   [1] LC_CTYPE=fr_FR.utf8       LC_NUMERIC=C
>   [3] LC_TIME=fr_FR.utf8        LC_COLLATE=fr_FR.utf8
>   [5] LC_MONETARY=C             LC_MESSAGES=fr_FR.utf8
>   [7] LC_PAPER=fr_FR.utf8       LC_NAME=C
>   [9] LC_ADDRESS=C              LC_TELEPHONE=C
> [11] LC_MEASUREMENT=fr_FR.utf8 LC_IDENTIFICATION=C
> 
> attached base packages:
> [1] stats     graphics  grDevices utils     datasets  methods   base
> 

-- 
Ben Bolker
Associate professor, Biology Dep't, Univ. of Florida
*** NEW E-MAIL ADDRESSES:
***   bbolker at gmail.com , bolker at math.mcmaster.ca
bolker at ufl.edu / people.biology.ufl.edu/bolker
GPG key: people.biology.ufl.edu/bolker/benbolker-publickey.asc

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 261 bytes
Desc: OpenPGP digital signature
URL:
<https://stat.ethz.ch/pipermail/r-devel/attachments/20100607/6c5fb127/attachment.bin>

Dear Eduardo

I have been testing!  Please try this experiment.

Create a pdf figure in R. Suppose you call it "myfig.pdf"

Then run
> embedFonts("myfig.pdf", out="newmyfig.pdf")
For me, that embeds fonts and solves the problem of symbols rendering
badly in the newmyfig.pdf.

After you become confident in this, it seems also to work to just
write the same pdf on itself.
> embedFonts("myfig.pdf")
does rewrite the original figure.

I don't know the implications of this for the file size of a pdf that
is created from Latex that incorporates a lot of pdf figures.  Will
test soon.

-- 
Paul E. Johnson
Professor, Political Science
1541 Lilac Lane, Room 504
University of Kansas

I sent the message accidentally to r-help, but I intended to follow
r-sig-debian because it is about the problem that Evince doesn't
render symbols properly.  it explains why I think the embedFonts
function in R is the "magic bullet"


---------- Forwarded message ----------
From: Paul Johnson <pauljohn32 at gmail.com>
Date: Mon, Jun 7, 2010 at 9:28 AM
Subject: Re: [R] Wrong symbol rendering in plots (Ubuntu)
To: "Eduardo J. Chica" <ejchica at gmail.com>
Cc: r-help at r-project.org


On Fri, Jun 4, 2010 at 12:01 PM, Eduardo J. Chica <ejchica at gmail.com>
wrote:
> Hi I am having problems with the rendering of scientific symbols (mu and
> degree) in my plots. Whenever I use these symbols they are rendered changed
> (mu is changed to the proportionality symbol and degree is changed to
> something resembling a gamma) in the X-device. If I make a pdf of the plot
> and open the file in Evince or Okular symbols are also rendered wrong,
> however if I open the file with Xpdf or Acroread they are rendered
> correctly.
>
> I did not have this problem before, it arose after I upgraded both R and my
> system (Ubuntu karmic koala to lucid lynx), so I can not tell for sure if
> the problem is R-related or Ubuntu related (I have posted in a Ubuntu forum
> also http://ubuntuforums.org/showthread.php?t=1325289).
>
> Please let me know if you have any idea of how to fix this, or if you can
> confirm this is not an R-related issue.
>
I run Ubuntu as well and I see the same problem when I view a pdf with
Evince, the default document viewer. ?In Adobe's Acorbat reader,
acroread, the symbols show correctly.

Here's an example pdf file:

?http://pj.freefaculty.org/linux/pm92.pdf

In Evince, instead of mu, you see the proportional to symbol.

In the Ubuntu bugtracker, this problem has been reported and they say
it has been fixed, but I still see it.


https://bugs.launchpad.net/ubuntu/+source/poppler/+bug/518236
http://art.ubuntuforums.org/showthread.php?p=9410094

I believe the problem is a mismatch between the fonts available to R
and to the various viewers. ?R does not embed all fonts in the pdf
files. ?Open a suspect pdf in Evince or some other viewer and check on
the fonts. In Evince, under the File/Properties menu, there's a font
indication.

None of the fonts in R pdfs are embedded. ?There is a way to embed
fonts after in R, (see ?embedFonts) but I can't understand the
documentation on how to do that.  But I'm testing it out.

I believe that the fonts installed in the system, combined with
whatever user config changes occur, make this a tough problem to
solve.

--
Paul E. Johnson
Professor, Political Science
1541 Lilac Lane, Room 504
University of Kansas



-- 
Paul E. Johnson
Professor, Political Science
1541 Lilac Lane, Room 504
University of Kansas

Hello, please excuse me if this is a repost, I think I got confused 
about how to reply. Anyway, I posted the message below about a day and a 
half ago through Nabble but have not been cleared until today. Here it 
goes again. Please bear with me I will get better with time.

----Start of message sent about two days ago---

Hello all, thank you very much for the replies and I am very sorry for 
being so late to come back to the discussion, I was on a field trip 
without internet access.

@ Ben

I have the problem any time I try to use the mu symbol or the degree 
symbol using expression() (the only way I know how to do it). Yvonnick's 
code is a good example, I have just added the mu and the degree symbol:

pdf("Test.pdf")
plot(1:10,xlab=expression(mu), ylab=expression(C*degree~pi))
dev.off()

Even if I don't make the pdf (i.e. just plot on the X11 device), like in 
Yvonnick's example, the degree symbol is still showing as that 
gamma-like symbol, pi is shown like an inequality symbol but mu is shown 
right (on the X11).

@ Erik and Yvonnick

I think Yvonnick's problem is the same I have (i.e. it is not restricted 
to pdf files, it also happens in the X11 device, jpg and png). Further, 
if I create a pdf in Ooo Writer that contains mu, pi or degree symbols 
and I export them as pdf, they are rendered just fine (similar to what 
Peter reported, not sure whether he has problems in the X11 device too), 
the problem is restricted to R-generated graphs.

I am going to try Erik's suggestion tonight (sorry I could not try it 
before posting, I am still catching up with work accumulated from my 
trip). However, would this fix the problem in rendering in the X11 
device? I also feel this might not be just a pdf-viewer issue, but I am 
an absolute novice.

    ----End of message sent about two days ago---

Since then, i tried Erik's suggestion, but I could not find  either 
'~/.fonts.conf' or '/etc/fonts/local.conf' in my system (at
least with
those exact names).

I also got a reply from Prof. Ripley (quote below), I am working on it 
now, but maybe more experienced users will be more efficient in using it 
than me.

"Your first para confirms it.  The problem is font selection by 
fontconfig, which R's X11 device and Evince etc use and Xpdf or Acroread 
do not.  See ?X11 for ways to debug this."

Thanks!

Eduardo

 > sessionInfo()
R version 2.11.1 (2010-05-31)
i486-pc-linux-gnu

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=C              LC_MESSAGES=en_US.UTF-8
  [7] LC_PAPER=en_US.UTF-8       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

	[[alternative HTML version deleted]]