| Re: A Simple IDL Manifesto [message #63681] |
Wed, 19 November 2008 06:28  |
Jeremy Bailin
Messages: 618
Registered: April 2008
|
Senior Member |
|
|
> 13.) Add the "dimension" positional parameter (as present in TOTAL) to
> MEAN, STDDEV and so forth.
Hear hear!
> 14.) make it possible that we can have the last dimension 1.
> Everyone who dislikes it can use reform. I can't remember how often this
> misbehaviour has created funny results (bugs).
I don't know... about half the time, I love how it works now, and the
other half the time (like when I was debugging last night) I swear
ferociously at it. In any case, that almost certainly falls under the
category of would-break-too-much-existing-code.
-Jeremy.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| Re: A Simple IDL Manifesto [message #63723 is a reply to message #63721] |
Mon, 17 November 2008 22:18  |
lbnc
Messages: 15
Registered: January 2005
|
Junior Member |
|
|
On 17 Nov, 17:50, David Fanning <n...@dfanning.com> wrote:
> Folks,
>
> On my way to work this morning, after I my German
> lesson gelearnt, I started thinking about simple
> things ITTVIS could do to improve the IDL experience
> for old and new users of IDL alike. My criteria for
> my list is that it had to be something a junior ITTVIS
> software engineer could accomplish in a lazy afternoon,
> while knocking back a couple of cold Brewskis.
>
> Here is the list:
>
> 1. Make the COLOR keyword default to 1 for the PostScript
> device. The last time I wanted two-color PostScript output
> was in 1964, and if I want it now I can draw in black
> and white colors.
>
> 2. Make the BITS_PER_PIXEL keyword for the PostScript
> device default to 8. Setting it to 4 probably made
> sense in the 18th century (I'm speaking metaphorically
> in computer time, of course), but it makes absolutely no
> sense today.
>
> 3. Get rid of the TRUE keyword on the TV command. If you
> don't know how to tell if an image is a 2D or a true-color
> image, I'm certain Liam can enlighten you.
>
> 4. Add a POSITION keyword to the TV command so you don't
> have to spend an afternoon figuring out who to overlay an
> image with some information.
>
> 5. Oh, the hell with it, ditch the TV command and incorporate
> TVIMAGE or IMGDISP in it's place. (Personally, I would vote for
> IMGDISP.) That way you could have axes and everything else, too.
>
> 6. Make square-bracket array subscripting the default. Craig
> will get used to it, believe me.
>
> 7. Make the NAN keyword default to 1 everywhere. Why not!?
>
> 8. I realize it is difficult to tell when you are staring at
> a piece of paper just ejected from the printer, but make the
> LANDscape PostScript output LANDscape and not SEAscape. You can
> tell, all the time, when you view the PS output in GhostView.
>
> 9. Make UNXI computers default to TRUECOLOR visuals instead of
> DIRECTCOLOR. Oh, wait... this is already done. Hurray! Scratch
> this one from the list.
>
> Cheers,
>
> David
> --
> David Fanning, Ph.D.
> Coyote's Guide to IDL Programming (www.dfanning.com)
> Sepore ma de ni thui. ("Perhaps thou speakest truth.")
Can I add:
10) Do *not* throw an error when I want to plot just NaNs. Instead,
print a warning and continue, maybe plotting only the axis. And David,
don't suggest I should get children to understand that there is
"always a hurdle to get over" - tried that, didn't help. Apparently
didn't help for you either...
Cheers
Lasse
|
|
|
|
|
|
|
|
|
|
| Re: A Simple IDL Manifesto [message #63850 is a reply to message #63783] |
Thu, 20 November 2008 08:42 |
Michael Galloy
Messages: 1114
Registered: April 2006
|
Senior Member |
|
|
On Nov 20, 2:11 am, Reimar Bauer <R.Ba...@fz-juelich.de> wrote:
> If one changes the code rules behaviour he can also write a migration
> tool which converts your old program into a better new program.
>
> But this wasn't also done by this company in the past, so we can assume
> it won't be in the future.
>
> And it is not unusual to do so. e.g. if the moinmoin wikisoftware
> project changes the wiki text syntax in a newer version we provide a
> migration tool for the old wiki text syntax on pages to the new syntax.
Yes, a similar change like this is happening for Python 3000 i.e.
Python 3.0. They are providing a py2to3 tool that will convert Python
2.6 to Python 3.0 code. This tool will do most of the grunt work, but
I believe some hand coding will still be necessary.
Backwards compatibility is a noteworthy goal, but the design of any
language will eventually show its age. I think eventually you have to
change things that turned out to be mistakes (hindsight is 20/20).
The things that I would change about the core language of IDL (not the
library) that would break backward compatibility would be:
1. get rid of that extra "blankity, blank" comma when calling a
procedure (the one right after the name of the procedure)
2. consistent handling of arrays with a last dimension of 1 (don't
remove dimensions for me, thank you)
3. allow arrays of length 0
4. make "compile_opt idl2, logical_predicate" the default
A conversion tool could probably do 1 and the "idl2" part of 4 pretty
easily. 2, 3, and the "logical_predicate" of 4 would be a bit harder
and probably require some overview by the developer.
Of course, this means that code written for this new "IDL 8" would not
work in previous versions (the tool would only convert from old to new
style). If the .sav file format didn't change, then at least "IDL
8" .sav files could be used in previous versions.
Another solution would more compile_opt flags, but I'm not sure what
should happen if a routine with the new compile_opt flag had an array
of length 0 and passed it to a routine without the new compile_opt
flag. And I'm getting tired of putting a compile_opt statement in
every routine I write.
Mike
--
www.michaelgalloy.com
Tech-X Corporation
Associate Research Scientist
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
