R devel - Feb 2011 - axTicks.Rd documentation bug

?axTicks says:

    usr: numeric vector of length four, defaulting to ?par("usr")?
          giving horizontal (?x?) and vertical (?y?) user coordinate
          limits.

 but this is not how the function is implemented -- in fact 'usr' should
be a vector of length two corresponding to the appropriate elements of
par("usr") [1:2 if side is 1 or 3, 3:4 if side is 2 or 4].

  A patch for src/library/graphics/man/axTicks.Rd against the latest SVN
is attached (I hope it makes it through).

  I also included an extended example of how to use axTicks without
reference to an existing plot in the logarithmic axis case: it took me
quite a bit of digging in documentation and source code to figure out
how to do this for myself, so I think it would be useful to others ...

   Ben Bolker


-------------- next part --------------
An embedded and charset-unspecified text was scrubbed...
Name: axTicks_patch.txt
URL:
<https://stat.ethz.ch/pipermail/r-devel/attachments/20110204/067c126d/attachment.txt>

>>>>> Ben Bolker <bbolker at gmail.com>
>>>>>     on Fri, 04 Feb 2011 10:44:12 -0500 writes:
    > ?axTicks says:

    >     usr: numeric vector of length four, defaulting to
    > ?par("usr")? giving horizontal (?x?) and vertical (?y?)
    > user coordinate limits.

    >  but this is not how the function is implemented -- in
    > fact 'usr' should be a vector of length two corresponding
    > to the appropriate elements of par("usr") [1:2 if side is
    > 1 or 3, 3:4 if side is 2 or 4].

    >   A patch for src/library/graphics/man/axTicks.Rd against
    > the latest SVN is attached (I hope it makes it through).

    >   I also included an extended example of how to use
    > axTicks without reference to an existing plot in the
    > logarithmic axis case: it took me quite a bit of digging
    > in documentation and source code to figure out how to do
    > this for myself, so I think it would be useful to others
    > ...

Thanks a lot, Ben!

I've also added the new example .. though slightly modified.
Notably I'm a tiny bit disappointed ;-) that you were not aware of
the existence of the  extendrange() function ...

Martin Maechler