| Re: IDLgrAxis expanding [message #15048] |
Thu, 15 April 1999 00:00 |
Pavel Romashkin
Messages: 166
Registered: April 1999
|
Senior Member |
|
|
Hi David,
Well, you didn't have to look up the CRANGE keyword in the on-line help - I
mentioned that keyword in my initial message! Doing taxes is really bad
for you, we all know that :-)
Thank you - the solution surely will be to re-normalize and rescale data
and axes into the range returned by CRANGE. This didn't come to my mind
since repeated rescaling is not something you think is totally normal. Why
in the world IDL's rounding expands axes PAST the viewplane rect. limits?!
So, you made smaller and simplier objects for plots? I should consider
attending your class - F. Collins is not so far from where I live
(Boulder). Unfortunately, mastering IDL is not quite what I am getting
paid for - papers that result from IDL use are. Travel and class fees are
not among the things most bosses like the best. Oh well, we'll see.
Thank you again and anjoy that beer!
Pavel
|
|
|
|
| Re: IDLgrAxis expanding [message #15049 is a reply to message #15048] |
Thu, 15 April 1999 00:00  |
davidf
Messages: 2866
Registered: September 1996
|
Senior Member |
|
|
Gwyn Fireman (fireman@mcst.gsfc.nasa.gov) writes:
> : But I am still confused about one thing: why "C"Range?
>
> I always thought it was "current" range.
> It may have been documented as such long ago, but I can't find it now.
No, I can't find it either, but I've already demonstrated
my sleuthing ability. Stein Vidar thinks "coordinate"
range. But either way, "current" or "coordinate", it comes
up empty in a search of the index. I've even been searching
on "B"Range and "D"Range without luck this morning. :-(
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting
Phone: 970-221-0438 E-Mail: davidf@dfanning.com
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Toll-Free IDL Book Orders: 1-888-461-0155
|
|
|
|
|
|
| Re: IDLgrAxis expanding [message #15055 is a reply to message #15048] |
Wed, 14 April 1999 00:00 |
davidf
Messages: 2866
Registered: September 1996
|
Senior Member |
|
|
I wrote after one too many beers:
> So here is how it is done. (It's a bit convoluted here,
> but it's a solution programmers will love.) I'm showing
> here just the calculations for the dependent data axis,
> but the other axes will be done in the same way:
Alright, I admit it. I was completely wrong. As Mark Hadfield
points out, you can use the CRANGE keyword to get the real,
autoscaled, range of the axis:
thisAxis->GetProperty, CRANGE=realRange
But this exercise wasn't completely pointless. I did come up
with a true "programmers" solution, and I did learn
a couple of things.
1. I learned something I didn't know about object axes and
Lord knows I've plenty of room to learn more.
2. I learned that it doesn't matter if you have looked up
the GetProperty method on an object 100 times and been
told to go to the INIT method 100 times. If you look
it up for the 101st time, you may find a keyword you
*can't* find in the INIT section of the manual.
And I re-learned that most of the information I want *IS*
in the IDL documentation, even if I don't know where to
find it. :-)
But I am still confused about one thing: why "C"Range?
What does that mean? I can understand "O"(output)Range,
or "R"(real)Range, or "A"(autoscaled)Range. But I just
don't get "C"Range. And I don't understand why I would
be expected to know that to find what I want in the index
of the on-line help. What I would prefer would be something
like this:
Axis
range
setting
obtaining the value of
I think even I could have found something like that. :-(
Cheers,
David
P.S. You still have to remember to apply the new axis
data range to the DATA, or your axis won't reflect your
true data values.
--
David Fanning, Ph.D.
Fanning Software Consulting
Phone: 970-221-0438 E-Mail: davidf@dfanning.com
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Toll-Free IDL Book Orders: 1-888-461-0155
|
|
|
|
| Re: IDLgrAxis expanding [message #15056 is a reply to message #15055] |
Wed, 14 April 1999 00:00 |
davidf
Messages: 2866
Registered: September 1996
|
Senior Member |
|
|
About a half hour ago I wrote this in response to a question
by Pavel Romashkin:
>> When an instance of IDLgrAxis is created without /EXACT keyword, it is
>> adjusted by IDL so that tick labels are nicely rounded. However, this
>> has a side effect of non-/EXACT axis extending past the limit of the
>> VIEWPLANE_RECTANGLE for the parent IDLgrView. Sometimes this expansion
>> is really large. Is there a way to make IDLgrAxis to round ticks but
>> lie inside the VIEWPLANE_RECTANGLE?
>
> If there is a way, it must be undocumented. :-(
Just after I finished this article (it being rather late) I said
to myself, "The hell with it, I'm going to have a beer". Of course,
I was out of beer. So on the drive to the liquor store the solution
came to me. :-)
I reasoned this way. *Someone* has to know the actual range of
the data, since it clearly gets drawn. Who is it who knows and
how can I get at that information? Using my best Sherlock
impersonation, I finally realized that it is the text object
that creates the labels that knows how long the axis is. So
how do I get at the labels? Humm. The STRINGS keyword! OK.
So here is how it is done. (It's a bit convoluted here,
but it's a solution programmers will love.) I'm showing
here just the calculations for the dependent data axis,
but the other axes will be done in the same way:
************************************************************
; Get the range of the data.
thisPlot->GetProperty, XRange=xrange, YRange=yrange
; Set up the scaling so that the axes for the plot and the
; plot data extends from 0->1 in the X and Y directions.
xs = Normalize(xrange)
ys = Normalize(yrange)
; Create the Y axis.
yAxis1 = Obj_New("IDLgrAxis", 1, Color=[255,255,0], Ticklen=0.025, $
Minor=4, Title=ytitle, Range=yrange, YCoord_conv=ys, $
Location=[0, 1000, 0])
; Get the text object containing the labels.
yAxis1->GetProperty, Ticktext=yAxisText
; Get the strings.
yAxisText->GetProperty, Strings=theseStrings
; Find the first and last of these strings. This is the REAL
; axis range.
nstrings = N_Elements(theseStrings)
minRange = Float(theseStrings[0])
maxRange = Float(theseStrings[nstrings-1])
newRange = [minRange, maxRange]
; Set the axis range to the REAL range.
yAxis1->SetProperty, Range=newRange
; Find new scaling values for this REAL range.
newYs = Normalize([minrange, maxrange])
; Re-scale both the axis AND the data into this new range.
yAxis1->SetProperty, Range=newRange, YCoord_Conv=newYs
thisPlot->SetProperty, YCoord_Conv=newYs
************************************************************
Works real good on my example program, even if it *is* a
bit of a hack. :-)
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting
Phone: 970-221-0438 E-Mail: davidf@dfanning.com
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Toll-Free IDL Book Orders: 1-888-461-0155
|
|
|
|
| Re: IDLgrAxis expanding [message #15058 is a reply to message #15055] |
Wed, 14 April 1999 00:00 |
davidf
Messages: 2866
Registered: September 1996
|
Senior Member |
|
|
Pavel Romashkin (promashkin@cmdl.noaa.gov) writes:
> When an instance of IDLgrAxis is created without /EXACT keyword, it is
> adjusted by IDL so that tick labels are nicely rounded. However, this
> has a side effect of non-/EXACT axis extending past the limit of the
> VIEWPLANE_RECTANGLE for the parent IDLgrView. Sometimes this expansion
> is really large. Is there a way to make IDLgrAxis to round ticks but
> lie inside the VIEWPLANE_RECTANGLE?
If there is a way, it must be undocumented. :-(
This is a maddening "feature" of the object axes. For example,
if you set the axis range from 0 to 12.6, IDL actually creates
an axis that has a range from 0 to 14. But if you ask the axis
what its range is:
thisAxis->GetProperty, Range=thisRange
It reports that its range is 0 to 12.6. Aaaughh!
Of course, it is impossible to scale an axis like this into
any sort of coordinate system, so you have two choices. You
can always use exact axis scaling (which is what I do because
I'm fairly anal and I like my axes pairs to be at nice right
angles to one another and where I put them, for God's sake).
Or, you can have IDL scale them haphazardly and have axes
end points sticking out every which way and never meeting
at a point.
You may have noticed that none of the IDL object graphic
examples ever use box-style axes. This is the reason why.
In object graphic's defense, it is quite easy (well, alright,
once you figure it out, it is quite easy) to write your programs
so that the user can drag the axes around and put them wherever
he or she likes. I suppose you could always "nudge" your axes
back together once they appeared if you were picky about
appearance, but this does seem a little over the top for
nice looking axes. Of course, this does nothing to solve
the problem of axes extending outside the "plot space".
> Any suggestions?
Yes, several. If axes are going to autoscale, at least
have them report the values they autoscale to. Failing this,
publish the axis autoscaling algorithms so that we can create
better guesses as to what our axes are really going to do.
But, frankly, if line plots are what you want to do, and
you want object-like properties, it probably makes more
sense to write your own "plot object" in direct graphics.
This has the advantage of simplicity and familiarity.
Plot objects are powerful and work nicely. And they
take 5 seconds to print rather than 5 minutes. Anyone
who has taken one of my object-oriented programming
classes could build one in about a third the time it would
take to build an object graphics plot. :-)
Cheers,
David
--
David Fanning, Ph.D.
Fanning Software Consulting
Phone: 970-221-0438 E-Mail: davidf@dfanning.com
Coyote's Guide to IDL Programming: http://www.dfanning.com/
Toll-Free IDL Book Orders: 1-888-461-0155
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
