[R] "Error in plot.window(...) : invalid 'xlim' value" from plot(...par(new = TRUE))

[Duncan Murdoch]

On 11-02-09 7:40 PM, David Wolfskill wrote:
> [New to the community; still in early part of R's learning curve.]

In the call below that has

plot(..., par(new=TRUE))

you don't name the parameter to which par(new=TRUE) will be bound, so 
the first available is chosen, which was apparently xlim.  That is, your 
code is the same as

plot(..., xlim=par(new=TRUE))

which is not what you meant.

Normally you put the par(new=TRUE) call after the plot() call, not 
within it, and then things will work:

plot(...)
par(new=TRUE)

This wasn't your question, but I generally advise against using 
par(new=TRUE), because it does more than some people think.  It throws 
away the scaling that was determined in the first plot, so you can't 
trust points in later plots to be at the locations that the axes show.
It is better to use the "add=TRUE" parameter to later plot functions if 
they support it, or (in your case) use

lines(...)

rather than plot(...) after the first call.

Duncan Murdoch

>
> Several months ago, I was requested to generate some graphs on a
> periodic basis.  Accordingly, I managed to figure out a way to do so,
> using a combination of Perl and R (in a FreeBSD environment).
>
> While I've needed to adjust a few things here and there, the general
> approach has been pretty solid , and the R part has had very little
> change perceived as necessary.
>
> I re-generate the graphs each Wednesday morning (as they show disk space
> utilization over time).
>
> This morning, one of the graphs -- the first that used plot() with
> "par(new = TRUE)" reported an error.  Re-doing the work interactively,
> I was able to re-create the problem.  Poking around with
> debug(plot.window) showed that while the ylim values make sense,
> the xlim values are ... of an unexpected type altogether.
>
> And when I re-tried the plot() invocation, but without the "par()"
> argument, the xlim passed to plot.window() was fine, and the plot was
> OK.
>
> The data look like:
>
>            Tag          Cap  mean_GB median_GB census
> 1  2010-02-03 540972711754 213.5250  154.1011   1061
> 2  2010-02-10 531711798208 169.4235  124.3809   1033
> 3  2010-02-17 544482444926 171.2384  130.0641   1052
> ...
> 52 2011-01-26 558625618800 195.2759  175.1277   1386
> 53 2011-02-02 571187807274 199.4980  170.4478   1406
> 54 2011-02-09 562448031310 197.5468  166.8976   1367
>
> I'm trying to generate a plot, then overlay 3 more plots on it -- which
> has been working for several months:
>
>> plot(Is_total$Tag, Is_total$Cap/1024/1024/1024, type = "l", col = "blue", ylim = c(0, max(Is_total$Cap/1024/1024, Is_total$census * 400)/1024), xlab = "Date", ylab = "Capacity(TB)", main = "Total Local Disk Space")
>> debug(plot.window)
>>   plot(Is_total$Tag, (Is_total$census * Is_total$mean)/1024, type = "l", col = "green", ylim = c(0, max(Is_total$Cap/1024/1024, Is_total$census * 400)/1024), ann = FALSE, axes = FALSE, par(new = TRUE))
> debugging in: plot.window(...)
> debug: .Internal(plot.window(xlim, ylim, log, asp, ...))
> Browse[2]>  ylim
> [1]   0.0000 549.2188
> Browse[2]>  xlim
> $new
> [1] FALSE
>
> Browse[2]>  c
> Error in plot.window(...) : invalid 'xlim' value
>> sessionInfo()
> R version 2.12.1 (2010-12-16)
> Platform: i386-portbld-freebsd8.2 (32-bit)
>
> locale:
> [1] C
>
> attached base packages:
> [1] stats     graphics  grDevices utils     datasets  methods   base
>>   plot(Is_total$Tag, (Is_total$census * Is_total$mean)/1024, type = "l", col = "green", ylim = c(0, max(Is_total$Cap/1024/1024, Is_total$census * 400)/1024), ann = FALSE, axes = FALSE)
> debugging in: plot.window(...)
> debug: .Internal(plot.window(xlim, ylim, log, asp, ...))
> Browse[2]>  ylim
> [1]   0.0000 549.2188
> Browse[2]>  xlim
> [1] 14643 15014
> Browse[2]>  c
> exiting from: plot.window(...)
>> q()
>
> This is running on my laptop:
> FreeBSD 8.2-PRERELEASE #99 r218478: Wed Feb  9 04:17:55 PST 2011     root at g1-222.catwhisker.org.:/common/S1/obj/usr/src/sys/CANARY
>
> (Yes, I track FreeBSD stable/8 on it daily.)
>
> I built R as a FreeBSD port; it is version 2.12.1.  (I update the
> installed ports daily, too.)
>
> I actually first noticed the behavior on my desktop machine at work; it
> is also running FreeBSD stable/8, but only updated as of 30 Jan 2011 --
> and the problem did not exist last week.  So I don't believe this is
> because of a change in FreeBSD.
>
> For now, I'm (also) open to circumventions, so I can avoid complaints
> about the graphs that weren't updated at work, but I'd like to resolve
> the issue.  (And yes, I fully understand that what needs to be fixed is
> my understanding of how to use the tools.)
>
> Thanks!
>
> Peace,
> david
>
>
>
> ______________________________________________
> R-help at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-help
> PLEASE do read the posting guide http://www.R-project.org/posting-guide.html
> and provide commented, minimal, self-contained, reproducible code.