[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: Plea for IDL 2000 (was: a plea for more reliable mathematical routines)





"Martin.Schultz@dkrz.de" wrote:
>   at least the quirks keep your attention focused on IDL ;-)
> (not on your scientific problem though ;-()  But to be constructive:
> how about a user-bug-fix initiative? A great many of IDL's routine
> are written in the IDL language, so everyone can go ahead and check them
> through (at least it's partly open source that way -- although RSI
> imposes a pretty stringent copyright notice on their routines). If
> RSI would be willing to cooperate here and perhaps set up a web site,
> ftp archive etc. where people could download the most recent versions
> of the IDL procedures, and upload their corrected versions (or submit
> them with some guarantee that they would appear in the archive with
> a reasonable response time) that could give them a bunch of programmers for
> free so to speak. To name an example: just last week I eliminated an
> annoying restriction in VELOVECT.PRO which is that it does not accept
> 2D arrays for X and Y and hence does not work for irregulariliy gridded
> data. Well, now it does;-) and I would be happy to contribute this fix.
> BUT OF COURSE: We need to be sure that our effort will be honored in that
> bug fixes will make it into the official distribution! Another example:
> wouldn't it be great to see MP_FIT appear in the official online help?

I am not satisfied with the procedure you are proposing. Firstly, I think
RSI should provide a state of the art math libary which is fully documented.
Secondly, the version changes in IDL are getting faster and faster and I
dont want to check their web site every week for new bug fixes and downloads.
At the end we tag every figure/table with: Based on IDL 6.1.122.a3;-)
I think there is no excuse today for a company selling an expensive(?)
software have problems with defined math. I went from C and Fortran to 
IDL some years ago under the assumption IDL would provide simultaneously
nice programming, great graphics and correct math. 

Theo

PS. It could be, however, that all the talk about the IDL math problems
is a hoax. Since the begin of this threat NOONE posted one of the putative
math problems. At this point I agree with Martin that we need a bug pool 
somewhere on the web.



-- 
----------------------------------------------
Dr. Theo Brauers
Institut fuer Atmosphaerische Chemie (ICG-3)
Forschungszentrum Juelich
52425 JUELICH, Germany
Tel. +49-2461-61-6646    Fax. +49-2461-61-5346
http://www.kfa-juelich.de/icg/icg3/MITARBEITER/th.brauers.html