axis handling - wishlist

[Martin Schultz <mgs(at)io.harvard.edu>]

Hello (especially to people from RSI if they are sneeking in),


   today I finally got around to take a closer look at the IDL axis
handling and spend some thoughts on what I would wish to see in a future
release of IDL. Maybe one or the other of you would like to comment or
contribute so that we could end up with an even better solution. Th
eordering does not express any ranking but is meant to facilitate
references.

   Basically I find that one can very well live with the general scheme
of !X, !Y, and !Z variables and all those hundreds of parameters that
can be passed to plotting routines. However, I think some tasks could be
greatly simplified if the behaviour of some of the elements of !X, !Y,
and !Z were slightly changed, and a few more items were added as well as
a few more keywords allowed to the plotting routines. Here are my
suggestions:

!X[YZ] variables:
-----------------
 (1)   !X[YZ] should contain an extra field with their COLOR (index).
       Additional options would be a field for TITLECOLOR and LABELCOLOR

 (2)   the !X[YZ].STYLE fields should have an additional NOLABEL flag
       which would have the same effect as TICKNAMES = " ", but is
       somewhat easier to use and remember. If TICKNAMES is set to " "
       this NOLABEL flag could automatically be set.

 (3)   !X[YZ].THICK = 0 (or -1) should draw no axis (it's kind of
       logical, or not?) 

 (4)   !X[YZ].TICKS = -1 should prevent tick marks (and minors)

 (5)   !X[YZ].TICKLEN = 0 should also prevent tick marks

 (6)   !X[YZ].GRIDSTYLE = -1 should be default, whereas a GRIDSTYLE ge 0
       would automatically set TICKLEN to 1 except if TICKLEN is
       explicitely given 

 (7)   !X[YZ] could sometimes use an additional OFFSET or SHIFT field
       which for !X would specify an offset in Y, and for !Y an offset
       in X (haven't thought about !Z). Together with the additional
       PLOT options I propose below, this would allow easily to draw
       plots with multiple (more than 4!) axis.

 (8)   two (or three) additional system variables should be 
       !XTOP and !YRIGHT (!Z???) to hold axis information on the
       secondary axis. Structure identical to !X[YZ]
 

additional keywords for PLOT etc.:
----------------------------------
 (9)   /XBOT, /XTOP, /YLEFT, /YRIGHT would tell PLOT to use (only) the 
       specified axis. Default would be having each pair identical
       (i.e. draw a box as it is now). If e.g. /XBOT is set, PLOT should
       only draw the lower X axis. A subsequent call with 
       PLOT,...,/XTOP,/NOERASE could then be used to add data to the 
       plot drawing and using the top axis only. I recognize that some
       thoughts would have to be spent on how to handle the "other" 
       axis, whether to simply overdraw the previous one or omit it.
       But since I am not paid by RSI ...

 (10)  XAXIS=XAXIS and YAXIS=YAXIS would be two keywords which allow
       passing of an axis structure to the plotting routine. In this
       case the routine(s) should retain the old values of !X[YZ] and
       use the given definitions instead. This way one could easily
       define a set of axis for various plots and then pass them along
       without having to extract all the keywords individually. If
       then e.g. the COLOR element of !X[YZ] were included, it would be
       possible to produce a plot with different colors for the data and
       the individual axis in one command. A subsequent PLOT without
       these keywords would then use the original !X[YZ] structures
       again.      

 (11)  If the /NODATA keyword is set, PLOT (or the other routines)
       should work even if no data is passed, i.e. create some dummy
       data within the routine or catch this error otherwise.

 (12)  It happens ;-) that the default position of TITLE does not lead
       to sophisticated results, or one wants to have multiple lines
       with various character sizes, etc. For these instances, it would 
       be very helpful to have an additional coordinate system which 
       would be similar to NORM except that 0 and 1 would specify the
       boundaries of the plotting window itself. This helps especially
       when annotating plots in a !P.MULTI environment. It would allow
       better use of subroutines which produce a complete plot on a 
       single page: they could more easily be embedded in "programs"
       which set up the !P.MULTI variable, open a device, and produce
       a number of plots. Currently one has to take special care and
       perform some weird calculations in order to place these labels
       correctly. A WINDOW coordinate system would allow a more
       consistent relative positioning. 
      

I tried to keep the changes as minimal as possible in order to ensure 
compatibility between this dream version and the current release.
If I am not mistaken, any "old" use of the plotting parameters should
still give the same results. The only incompatibility might arise if
someone uses a copy of say !X, re-creates this like 
     !MYAXIS={ ... }
and tries to copy it back into !X. That would obviously miss the
additional tags that I proposed. 



Looking forward to some discussion,
Martin.



-------------------------------------------------------------------
Dr. Martin Schultz                   
Department for Earth&Planetary Sciences, Harvard University
186 Pierce Hall, 29 Oxford St., Cambridge, MA-02138, USA

phone: (617)-496-8318
fax  : (617)-495-4551

e-mail: mgs@io.harvard.edu
IDL-homepage: http://www-as.harvard.edu/people/staff/mgs/idl/
-------------------------------------------------------------------