[Tikzdevice-bugs] [R-Forge] tikzDevice 0.5.1

[Kurt Hornik]

>>>>> Charlie Sharpsteen writes:

> On Tue, Oct 12, 2010 at 12:03 AM, Kurt Hornik <Kurt.Hornik at wu.ac.at> wrote:
>> 
>> I checked again: I had the same problem with the old version, and in
>> fact, looking at the package check results page at
>> http://cran.r-project.org/web/checks/check_results_tikzDevice.html you
>> will see that we get the same problem on Fedora, Solaris, Windows and
>> MacOS X: so I guess there is nothing Debian specific to reproduce.
>> 
>> I'll attach my log file below, perhaps you can spot something or spot
>> where you might be using private include files in your case which make
>> things work without problems.
>> 
>> I'll put the new version on CRAN for now.
>> 
>> Best
>> -k

> Hi Kurt,

> I never looked into the recent NOTES on the CRAN check; I just assumed
> that it referred to the first note:

> checking R code for possible problems ... NOTE
> queryMetricsDictionary: no visible binding for global variable
> ‘dictionary’

> This has plagued us for a while--- we were never able to find a
> resolution in the manuals, the mailing lists or by hacking around.
> Consequently, we just accepted there was a NOTE... now it appears
> there is more than one.

> However, in any case it appears that only the checks run against
> prerelease versions are failing.  It looks like this is R 2.13.0 and
> 2.12.0-alpha or 2.12.0 RC.  Strangely r-prerel-windows-ix86+x86_64 is
> passing.  All checks run agains 2.11.1 are passing for 0.5.0.

> Looks like there may be changes in 2.12.0+ outside of the addition of
> polypath() that we need to worry about.  I will wire up a build of the
> release candidate and see if it shows me anything.

Confirmed on my end: if I use 2.11.1, the problem goes away (also for
0.5.1).

Did you already try 2.12.0?

Best
-k



> -Charlie