Re: Just some thoughts

From: Mary Haley <haley_at_nyahnyahspammersnyahnyah>
Date: Wed, 29 Sep 2004 10:10:46 -0600 (MDT)

> This post addresses some issues that I have discussed with more than a
> few developers of graphics routines.
> Examples are very good as an introduction to a graphics system. They
> show capabilities and give a start to the basics.
> However, it is essential to have the next step. The examples inevitably
> lead the user to say ' That is just what I want to do but
> with ......something different.' There must be another stage of
> documentation or cookbook which discuss the most common variations
> on the examples themes. In my case, I saw the contours over maps and
> wanted to have the lat,lon ticks labeled at specific intervals.
> The next step after the examples are lists of hundreds of resources
> addressing the tick marks and labels and map plots, and the complex
> interaction between the them.
> I have used GMT (Generic Mapping Tools) for many years. It has a
> cookbook and very well thought out man pages. The path from the
> examples to custom plots is clearly delineated by the man pages. GMT
> benefits in this sense by having less capability than NCARG.
> I feel that there needs to be some intermediate description of 'best
> sellers' of 'most used' resources. Some gradual progression of
> complexity rather than a abrupt cliff.
> From what I can gather, this is what lead to the GSUN aspect of NCL.
> Exactly analogous to pyngl, ncl exposed the complexity of the HLUs and
> users recoiled in confusion. GSUN made life simpler. The parallels
> between ncl and pyngl are quite close, except python's syntax is much
> clearer and the language is far more powerful with tremendous support
> from its standard libraries (for example there is built in RGB->HSV
> support).
> So - what I see is that pyngl needs to reproduce gsun - this is an
> unfortunate duplication of effort but the experience garnered by the
> previous effort should speed things along.
> -Jim

Hi Jim,

Thanks for your thoughts on the examples. I agree with you 100%,
because it's been the mass number of examples that has helped the
users the most, rather than wading through the hundreds of resources
and trying to figure out which ones to set.

One of our developers has been talking about the idea of a
NCL-to-PyNGL converter that you could run on an NCL script to create a
PyNGL script. Obviously, there are enough differences between NCL and
Python that you will probably still have to convert by hand, but it
certainly gives us a head start in converting our examples.

This will become high on our priority list, to provide more PyNGL
examples. We will also try to do it in such a fashion where we
start off with more simple examples, and work our way up to the
complicated ones.

The examples that are currently part of "pynglex" were done more for
the "gee whiz" factor. We were preparing a talk, and had already
created these examples to put in our slides, so we just decided
to make them part of "pynglex". We now need to work our way back
down to some simpler examples. :-)


pyngl-talk mailing list
Received on Wed Sep 29 2004 - 10:11:17 MDT

This archive was generated by hypermail 2.2.0 : Thu Jan 19 2006 - 21:30:16 MST