On Thursday, 5 March 2015 17:00:00 UTC+1, Páll Haraldsson wrote:
>
> In astronomy MATLAB is definitely not standard.
>>
>
> He's not doing astrobiology, it's more down to earth :) I was just
> repeating what he said,
>
I know. I was replying to his comment. I know it wasn't from you.
> not because I believe it applies to the whole scientific world (he
> probably meant only in the systems biology field, if that); I thought
> (proprietary) MATLAB code shouldn't a requirement for any journal
> (publisher requirements, or familiarity for readers, besides Julia is close
> enough..).
>
I would be shocked if any journal required MATLAB, or any specific tool. I
can't imagine how that could possibly make sense to anyone. A lot of
important discoveries might not even require a lot of computation at all
(e.g. imagine some experiment with an interesting result).
> Leaving aside the compiled languages (Fortran and C) the most common
>> language in astronomy is definitely IDL.
>>
>
> I haven't really looked much into IDL (and probably will not), still
> interesting that this language from 1977 is still in use.
>
It's a terrible language. It is only used because people tend to learn
whatever their supervisor uses, and a lot of important tools are written in
that language. But the language itself is horrid. The syntax is ugly, the
API is gratuitously obtuse, and the license fee is astronomical (pun
intended). I consider myself lucky. I do theory work and that means I have
more freedom to create my own tools if I want to.
> Since similar to MATLAB, I assume IDL also likely could get killed by
> Julia(?).
>
Hope springs eternal. I hope you are right. I truly hope that IDL suffers a
sudden but painful death. But there is so much momentum behind it that I
very much doubt that IDL will go away any time soon.
> "IDL Save/Restore files embody unpublished proprietary information about
> the IDL program. Reverse engineering of this file is therefore forbidden
> under the terms of the IDL End User License Agreement (IDL EULA)."
>
Yup. And the license costs several thousand dollars. Yeah... IDL makes
MATLAB seem friendly and open. At least MATLAB files are just HDF5 files.
MATLAB has the sense to follow a published open standard to save stuff,
which is why Julia can read it. I fully expect that IDL files are nothing
more than a memory dump or something similarly nasty.
> Julia is definitely a young language, and I consider myself an early
>> adopter. For example, for plotting I need to use Python's Matplotlib (via
>> the PyPlot package) because the native Julia plotting packages are simply
>> not mature enough.
>>
>
> I assume there is nothing wrong with that, while you could re-implement a
> lot of stuff, I see it as a plus for Julia how much has been reused with
> wrappers and how easy it is.
>
I think it's fine. Julia made me a fan of Matplotlib. You see, I don't like
Python, but I think Matplotlib has a lot going for it. Therefore, I
imagine/hope that some Pythonistas might be converted to Julia if they know
that they can keep Matplotlib.
> I used to be a big Perl enthusiast. Today Julia has replaced Perl as my
>> default "go-to" language for regular daily work. I hardly use Perl anymore.
>> When I need to make a quick script to grab stuff from a text file I usually
>> write a shell script with grep, sed and awk.
>>
>
> I still use/d those little languages, (I see at least one Julia developer
> do too (along with Julia); in the Fortran transpiler.., stage 1). I guess
> there's nothing wrong with that, old habits die hard. I never graduated to
> Perl, and while am not too familiar with it I'm not sure if it's more
> natural for these things than Python and/or Julia. I suppose none of these
> are a perfect replacement for the dataflow idea of pipes (not that you
> can't use pipes with Julia).
>
It is not just that old habits die hard. No language can be the best at
everything. If you want to "grep", then grep is definitely better than a
Julia script. I could write a Julia script to do ... "for f in `find .
-name '*.in'`; do grep x|egrep -v '(y|z)'|gawk '{print $2}'| sed 's/a/A/';
done" ... but it would take a lot more time to write the same in Julia.
Cheers,
Daniel.
