GNU Data Language | Hacker News Reader

GNU Data Language

github.com

78 pointstosh30 comments

I've said it before and I'll say it a thousand times more: when showing a language...

Show. Me. The Code.

I had to click through 5(!) links before I saw any code and those links were not terribly intuitive. No "getting started", no "tutorial", nothing. It seems like an interesting language but it baffles me how many language designers and implementers seem to hide the very code that they aim to use.

Here is the first bit of code I could find: https://www.l3harrisgeospatial.com/docs/bubbleplots.html

You'd only be looking at this because you already know you need to use IDL and want an open-source implementation. Unless you have a hard requirement to use it (like your data is in an IDL data format, or you rely on IDL libraries) there is no reason you would choose to over R, Python, or Julia.

Is it really a language, though? It seems simply like an implementation of an existing language. If I make a new C compiler, you probably won't ask me for example inputs for it.

Yes but if you go to the website for the original language then 5 links shrinks to 4. That's still 4 too many

If you remove the loitering '$' at line ends, and replace ';' for comments with '#,' this would pass as valid Python.

yes, prose for the enlightened. That annoys me, unenlightened as I am, and keeps me out.

its actually pretty decent syntax. I wonder why they wouldnt put it more front and center?

share the resulting link!

Sure thing. Edited my original post

I worked with IDL 20 years ago. GDL at the time was somewhat different and the difference was great enough to prevent me from doing the entire thing in GDL. Great to see they made the move to be compatible with IDL. It has a great REPL. Easy to program. Could have been the python we never knew we wanted.

I used it once for processing matrices (~20ya). I was very new to programming, and it completely threw me off as I wasn't prepared for a lot of its concepts.

Here is a language sample from Rosetta code:

https://rosettacode.org/wiki/Mandelbrot_set#IDL

Is this actually associated with the GNU project? The GNU folks aren't fond of GitHub. [0]

[0] https://www.gnu.org/software/repo-criteria-evaluation.en.htm...

The Wikipedia page for GDL states that it isn't an official GNU package. [0] I couldn't find an explanation as to why it is named such.

[0] https://en.wikipedia.org/wiki/GNU_Data_Language

Well they can probably get in trouble for branding as GNU, even if they disclaim it somewhere else. GNU is trademarked, for one. https://trademarks.justia.com/853/80/gnu-85380218.html

Would be nice to have some language samples in the README.

Yeah I always hate it when projects do that. It should be the first thing in your README. Instead I can’t even find an example clicking through the docs. I guess I’ll pass for now

It's funny for me to see all these references to 'IDL' describing "Interactive Data Language". I had only associated the term IDL with 'Interface Definition Language' [1].

[1] https://en.wikipedia.org/wiki/Interface_description_language

https://en.wikipedia.org/wiki/IDL_(programming_language) is the one you're looking for :)

It's a popular (and frankly terrible) language used in astronomy and astrophysics. My first (semi-professional) programming job was a high school internship at NASA AMES writing IDL to process data from the Cassini probe orbiting Saturn. At that time you had to use a clunky IDE for development and the whole toolchain was proprietary.

You never had to use their IDE, and in fact the IDE didn't exist for the first 20 years or so. Being an interpretive language, no toolchain involved. You just write code in any editor and run the IDL interpreter, just as you do with perl or python. Code libraries were discovered via your IDL_PATH environment, similar to how perl and python work.

What are the advantages of GNU Data Language over established alternatives such as R, Python/NumPy, or Octave?

Main advantage would be running existing IDL programs without the proprietary license. For any new development the alternatives you mention have a larger userbase.

Octave is also open source. It's GNU Octave, in fact

GNU Octave is intended as a non-proprietary replacement for MATLAB not for IDL, so, as another poster has written above, both these 2 GNU projects have the purpose of replacing certain proprietary tools, but they are intended for different user bases, who were previously familiar with those 2 different proprietary tools.

IDL is a much older established language dating back to the '70s. So GDL would appeal to anyone needing to run or develop IDL code but wishing to avoid paying the license fee.

I remember using IDL to analyze atmospheric observations made by satellites back in 1991.

And not a single hello world example anywhere. This is a common theme, unfortunately.

These have to be some of the worst "clearly had time and money invested in them, but are totally shit" docs I've seen in a long while. No code samples, takes ~7 seconds to load each page and the "Concepts" page (a natural starting point) is blank [1].

Maybe "Explore IDL routines" might have some interesting stuff? Nope [2]. The most concise and useful page I've found seems to document the "IDL _help_ Syntax" [3].

I can't get over how poor these docs are.

1. https://www.l3harrisgeospatial.com/docs/idl_concepts.html

2. https://www.l3harrisgeospatial.com/docs/routines-1.html

3. https://www.l3harrisgeospatial.com/docs/idl_syntax.html

Reply on news.ycombinator.com