R devel - Jul 2004 - fedora core 2 + bitmap() + ghostscript (PR#7062)

Hi all,
	A bug popped up recently when I upgraded to fedora core 2 (from core
1).  Whenever I used the bitmap() function, ghostscript complained that
it couldn't handle png or tiff devices.  This was NOT a bug with R.  It
was with the version of ghostscript shipped with core 2.  Going to
rawhide and getting ghostscript 7.07-29 (core 2 comes with 7.07-25)
solved this problem.

Unfortunately, I have no output to demonstrate this problem as I FINALLY
managed to fix it, and there's no way I going back :)

Thanks  Piers


-- 
/*****************************/
Dr Piers Dunstan
Marine Research
CSIRO
Hobart, Tasmania

Ph: 613 6232 5382
Email: Piers.Dunstan@csiro.au

/************************************/

On Tue, 2004-07-06 at 19:01, Piers.Dunstan@csiro.au
wrote:
> Hi all,
> 	A bug popped up recently when I upgraded to fedora core 2 (from core
> 1).  Whenever I used the bitmap() function, ghostscript complained that
> it couldn't handle png or tiff devices.  This was NOT a bug with R.  It
> was with the version of ghostscript shipped with core 2.  Going to
> rawhide and getting ghostscript 7.07-29 (core 2 comes with 7.07-25)
> solved this problem.
> 
> Unfortunately, I have no output to demonstrate this problem as I FINALLY
> managed to fix it, and there's no way I going back :)
> 
> Thanks  Piers

Here is some example output, which is a problem with the default version
of GS in FC2:
> bitmap("test.png")
> DEBUG: no psname was provided.
DEBUG: no psname was provided.
DEBUG: no psname was provided.
plot(1:10)
> dev.off()
null device
          1

However, despite the msgs, the PNG file _is_ created with the plot.

I have noted the above problems when opening postscript files using gv,
where the above DEBUG msg appears in a pop up window. You have to clear
the pop up to get the file to display. This has been a problem since
moving to FC2.

Note that I installed the updated '-29' rpms from fedora development and
re-ran the sequence of statements _during the same R session_ and the
second sequence below is the result, with the DEBUG msgs gone:
> bitmap("test.png")
> plot(1:10)
> dev.off()
null device
          1


This should not have been reported however to R-bugs, since as you note
above, it is not a bug in R. An FYI "heads up" msg to r-devel and/or
r-help would have been entirely reasonable though.

Marc Schwartz