On 10 September 2008 at 17:11, Christian Hudon wrote:
| Debian Bug Tracking System wrote:
| > This is an automatic notification regarding your Bug report
| > which was filed against the r-cran-cairodevice package:
| >
| > #498527: Cairo should be default output device in R instead of X11 when
r-cran-cairodevice installed
| >
| > It has been closed by Dirk Eddelbuettel <[EMAIL PROTECTED]>.
| >
| >
| > Thanks for the bug report.
| >
| >
| > | This would probably requires cooperation with the base R packages...
| > | feel free to reassign to a more appropriate R package if this is the case.
| >
| > ... that is unlikely for contributed packages. A but like lower priorities
| > in Debian -- 'base R' just won't depend on contributed packaged. Sometimes
| > the contributed package morph into base R but that is R.
| >
| > There is nothing I can do here for you. Part one of what you suggest may be
| > a local admin issue, part two is clearly something you'd need to argue with
| > the cairoDevice maintainers and on r-devel.
| >
| Maybe I should have been more explicit about what I had in mind. I
| wasn't suggesting having base R depend on r-cran-cairodevice. The two
I understand but ...
| ways of doing this that I can see are:
|
| 1. If there's a directory of .r files that are read at startup by R,
| like /etc/emacs/site-start.d for emacs, etc., then r-cran-cairodevice
| could add a file in there that would change the default output device
| from X11 to Cairo.
|
| Having the equivalent of /etc/emacs/site-start.d for the use of Debian R
| packages would require a change to the R core package, but it should be
| doable since R is a general programming language. And this is something
| that might be useful for other R Debian package after.
... I as maintainer have 'almost always' shied away from making 'our' Debian
R non-standard. There are little things (paperconf, say) I am willing to do;
what you suggest is a change in basic functionality that would surprise as
many users as it may please.
Please discuss upstream, this is not a Debian bug in my book. You want
different behaviour _upstream_.
| 2. Another option is to simply have the Debian startup code for R (in
| r-base-core, I guess) enable the Cairo output device if it sees that the
| cairoDevice library is installed, otherwise use X11.
Which is why I said 'local admin issue'. Experiment with your site.profile;
this is what it exists.
| Both ways, core R doesn't depend on r-cran-cairodevice, but users that
| install the package automatically get good-looking graphic output (like
| people have by default with R on Mac at least).
|
| Please reopen this bug report, at least to discuss a bit... that is
| unless you find both suggestions completely atrocious.
They are surely not atrocious, but they are suggestions of the user of one
among now almost 1500 packages at CRAN. We can't win if we single them out
this way. Too many potential special issues. Everybody will have a pet
peeve.
Also, did you actually realise that 'cairo' is now a capability of R itself?
Call 'capabilities()' or 'capabilities("cairo")' in R 2.7.2:
> capabilities("cairo")
cairo
TRUE
>
There were some speed issue with Cairo in R 2.7.0, IIRC.
To sum up, R strives to be consistent across platform and installations. I
will not be in favour of deviating from that path.
I am sorry of this strikes you as dogmatic. I am actually trying to be
constructive, albeit also in the spirit of R.
Cheers, Dirk
--
Three out of two people have difficulties with fractions.
--
To UNSUBSCRIBE, email to [EMAIL PROTECTED]
with a subject of "unsubscribe". Trouble? Contact [EMAIL PROTECTED]
