Skip to content

par help wrong: xlog & ylog (PR#497)

2 messages · Martin Maechler, Thomas Lumley

Original
Martin Maechler
# 

        
Bill> On 21 Mar 2000, Peter Dalgaard BSA wrote:

        
>> wsi@gcal.ac.uk writes:
    >> 
    >> > Full_Name: 
    >> > Version: 1.0.0
    >> > OS: linux
    >> > Submission from: (NULL) (193.62.250.209)
    >> > 
    >> > 
    >> >     
    >> > 
    >> > ?par lists and describes xlog and ylog, though these are
    >> > obsolete. It does not list log. I found out about log from
    >> > ?plot.defaults--so that help is OK.
    >> 
    >> Not a bug, not obsolete, and works as documented:
    Bill> True. True. Not documented in plot help. It is documented in plot.default.
    Bill> Why not in par, where xlog and ylog are? I find it confusing.

    >> 
    >> > plot(1,log="y")
    >> > par("ylog")    
    >> [1] TRUE

log is *not* a valid  par(.) argument, what makes you think it should be
documented there ?

  > par(log = "x")
  Warning message: 
  parameter "log" can't be set in: par(args) 


It *is* a valid argument to most(?) plot methods;
plot is the generic function, and the idea is that it only documents what
*ALL* plot methods have in common.  This is the reason that "log = .." is
documented with the default  plot method, plot.default.

You might have a point:  We could (should we?) consider requiring a 
"log =" argument for *ALL* plot methods, and then have it 
documented and possibly even among the listed arguements (the "non ..." ones)
of the generic  plot.

Opinions?

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._

  
  


  


    

    

      
      

        


  

        

      Thomas Lumley
    

    

      
      #
    

  


  

      
On Tue, 21 Mar 2000 maechler@stat.math.ethz.ch wrote:

        

            

      
      
      
    

        
Aaagh! No!  What about plots that intrinsically contain zero x or y
values, such as mortality plots (ie upside-down survival plots) or
Receiver Operating Characteristic curves?

Going further, some plot methods have a ... option but don't pass it to
plot. If the method uses plot to set up axes and then calls points() or
lines() to handle the actual drawing it may make sense to pass ... to
points() or lines().  These functions don't understand log=, so it would
be necessary to have an explicit log= option for the method, even if it
was completely inappropriate.

	-thomas

-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-.-
r-devel mailing list -- Read http://www.ci.tuwien.ac.at/~hornik/R/R-FAQ.html
Send "info", "help", or "[un]subscribe"
(in the "body", not the subject !)  To: r-devel-request@stat.math.ethz.ch
_._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._._