[R] cex.lab ignored in plot.zoo for multiple plots

[David Winsemius]

On Sep 11, 2012, at 7:48 AM, Gabor Grothendieck wrote:

> On Wed, Sep 5, 2012 at 11:06 AM, Nicolai Schneider
> <rstatistics.user at gmail.com> wrote:
>> Hello everyone,
>> 
>> a problem with the plot.zoo function. In the parameters of the function,
>> cex.lab is ignored. I tried to reduce the size of the yaxis labels by at
>> least 50%.
>> 
>> ------------------
>> 
>> Example:
>> 
>> sample <- as.zoo(EuStockMarkets)
>> 
>> par(las=1)
>> 
>> plot.zoo(sample, plot.type="multiple", main="Time Series", xlab="Date",
>> yaxt="n", cex.lab=0.5, xaxs="i")
>> # Try playing with different values for cex.lab
>> 
>> par(las=0)
>> 
> 
> Internally it uses mtext to write that text and mtext uses cex which
> is already taken.  Here is a workaround:
> 
> library(proto)
> p <- proto(plot.zoo = plot.zoo, mtext = function(...)
> graphics::mtext(..., cex = .5))
> with(p, plot.zoo)(sample)

So that creates an R environment where you are creating an unmodified instance of the plot.zoo function as well as installing a genetically-modified instance of mtext which has a redefinition of the parameter list to graphics::mtext()? (So using `(... , arg=newval)` is an acceptable way to over-rule a parameter list assignment? It certainly seems to be.) 

So when you access access that p-instance of plot.zoo(), it finds this new "genetically-modified" instance of mtext() because it is in the same environment?

And the reason it doesn't work to just redefine mtext (I tried) is that it would be in an environment that plot.zoo would not examine first when called from the command line?

-- 

David Winsemius, MD
Alameda, CA, USA