Re: "unset ytics" in gnuplot 4.2.3 mirrored tics drawn

[Shai Ayal]

On Mon, Mar 31, 2008 at 11:23 PM, John W. Eaton <address@hidden> wrote:
> On 31-Mar-2008, David Bateman wrote:
>
>  | In Octave the colorbar is part of the axis to which it is associated
>  | rather than a separate entity. So to get the feature you want we have
>  | two choices. We could add additional properties to the axis in addition
>  | to the __colorbar__ property to allow the setting of the colorbar tick
>  | properties.
>
>  I'd prefer to avoid doing that because I think it would be much better
>  if we avoid filling the properties database with entries that are
>  specific to individual backends.
>
>  For example, I added the key, keybox, and keypos properties because I
>  saw the legend as essential but did not have time to try to implement
>  it in a compatible way.  But eventually, we will need to do that,
>  (though I suspect we may never actually end up doing it for the
>  gnuplot backend if the OpenGL backend becomes the default).
>
>
>  | Or we could implement the colorbar as a separate set of axes
>  | with a callback to the axis it is associated with to set the color. The
>  | first would be relatively easy to implement, but means that the setting
>  | of the colorbar tick properties would not be matlab compatible. However
>  | the second seems quite tricky to get right and uses parts of the code
>  | that Shai and Michael are modifying. So I'm not sure of the best
>  | approach to get the behavior you are after.
>
>  I'd say wait until after the new graphics code is merged.
>
>  Perhaps we should have a plan for doing that?  I'd be happy to merge
>  any time if it is mostly working.  For now (and probably through the
>  3.1 release series) we could leave the gnuplot backend as the default
>  but allow switching to the OpenGL backend for testing purposes.

Merging now might be premature -- we are still missing text support.
This is the only big feature still missing from the OpenGL backend.

Shai