Re: Bugs in new key placement code

[Ethan A M. <merritt@u.washington.edu>]

On Sunday 05 February 2006 05:21 pm, you wrote:
> Alright, I've had a look at this issue.  
> I'm not so sure these are bugs in the new key placement code.  
> Rather, they are "bugs" in the old code, especially the 3d plots...  

Huh?  I think you are not focusing on the same issue
that I am.  Or possibly you are not even seeing the problem.
I'm not talking about the margin spacing at all.
Yeah, it's imperfect, but I don't care so much about that.
What I'm pointing out as a bug is ...


> >   set key box title "Foo"
> >   splot 1,2,3,4,5
> > 
> > Everything is OK so far.  Now try:
> > 
> >   set key rmargin
> >   replot
> > 
> > That's not what I expected, but let's proceed.
> > 
> >   set key tmargin
> >   replot
> > 
> > Huh?  The box is no longer correct.

By which I mean that the key box got smaller.
It was originally the correct size to contain the
set of titles.  But for some reason this command
causes it to shrink vertically, so the final title
now appears below the key box. That's got to be
a bug, pure and simple.  Once we figure out how big
the box needs to be, it should be drawn at that size
no matter where on the graph window it sits.

  
> And now it is the same sort of thing with margin key placement in 3d. 

I'm not worrying overly much about the placement.
It's the fact that the *size* is not calculated consistently.
Most of the time it is correct, but some key placement options
seem to break the size calculation.

> Because my personal feeling is that 2d/3d should be laid out the same in 
> terms of margins, so that code should be unified.

I disagree, but not strongly.  
Anyhow, that's not what I'm complaining about.

Since the time I originally sent this message, the centering code
seems to have been fixed.  Intentionally or unintentionally - I don't know.
But the incorrect centering I pointed to originally is no longer there
in current cvs.

> I'm not offering or committing to do any of this right now in light of
> the conversation of the last month regarding the future of gnuplot.  

Oh?  I didn't think much new came out of that discussion.
We're still in exactly the state we always were.
Except that we are now, I hope, much closer to a 4.2 release :-)

[gotta go back to emptying out that bug list]

-- 
Ethan A Merritt
Biomolecular Structure Center
University of Washington, Seattle 98195-7742