[Rd] R-2.8.0 : get platform device with get(getOption("device"))

[Prof Brian Ripley]

If you set, say

options(device="pdf")

you get the old behaviour.  You really cannot assume that session settings 
will be unchanged unless you set them yourself.

On Fri, 9 May 2008, Paul Roebuck wrote:

> On Thu, 8 May 2008, Peter Dalgaard wrote:
>
>> Prof Brian Ripley wrote:
>>> Here is the NEWS item, for 2.5.0
>>>
>>>     o    options(device = ) now accepts a function object as well as
>>>     the name of a function.
>>>
>>> The construct in the subject line has been broken ever since.
>
> Yet the change produced no warnings when made, so apparently
> everyone else missed it!

It did, if you set that option.

>> OK, but that entry was obviously hard to spot even when looking for it!
>>>
>>> On Thu, 8 May 2008, Peter Dalgaard wrote:
>>>
>>>> Prof Brian Ripley wrote:
>>>>> Please do study the posting guide.  This was covered recently:
>>>>>
>>>>> https://stat.ethz.ch/pipermail/r-devel/2008-April/049319.html
>>>> Nevertheless, perhaps an explicit NEWS entry could be warranted,
>>>> something like
>>>>
>>>>  o    The "device" option can now be a function, and is one in some
>>>> standard setups. Consequentially, get(getOption("device")) will fail;
>>>> programmers should usually use dev.new() instead.
>>>>
>> So...
>>
>> o    The "device" option now is a function in some standard setups.
>> Consequentially, get(getOption("device")) will fail; programmers should
>> usually use dev.new() instead. (Assuming that the device is a character
>> string has been incorrect since 2.5.0.)
>>
>
> Prof Brian Ripley noted [in another recent message] that
> it was also wrong to assume ANY arguments for dev.new
> method. If we are only concerned with interactive screen
> devices (i.e., x11, windows, quartz), would it be safe to
> come up with a method that can take basic dimension arguments
> (width & height)? Suggestions for an R-Core approved method
> of doing so?
>
> We use this functionality every day for creating multiple
> onscreen windows with specified dimensions.
>
> While I understand things have to change to evolve the
> codebase, this change broke the code supporting countless
> papers and analyses here in our department, without warning
> until now.
>
> ----------------------------------------------------------
> SIGSIG -- signature too long (core dumped)
>
> ______________________________________________
> R-devel at r-project.org mailing list
> https://stat.ethz.ch/mailman/listinfo/r-devel
>

-- 
Brian D. Ripley,                  ripley at stats.ox.ac.uk
Professor of Applied Statistics,  http://www.stats.ox.ac.uk/~ripley/
University of Oxford,             Tel:  +44 1865 272861 (self)
1 South Parks Road,                     +44 1865 272866 (PA)
Oxford OX1 3TG, UK                Fax:  +44 1865 272595