From: "Ara.T.Howard" <Ara.T.Howard@...>
Date: 2005-10-26T23:56:08+09:00
Subject: Re: narray and the stdlib

On Wed, 26 Oct 2005, Cameron McBride wrote:

> On 10/25/05, Phil Tomson <ptkwt@aracnet.com> wrote:
>
>> Personally, I'd really like it if it were in the standard lib because then
>> I wouldn't have to install it on all the different machines I want to use
>> it on (Currently running simulations on a Linux box, two Solaris boxen,
>> and OSX).
>
> Yeah, me too.
>
> Along this line, I thought it would be handy if code written with
> NArray could be shared with others on platforms less inclined to
> compile extensions.  My preferred platforms seem to always have a
> compiler in the path, but I hear this is not always the case.  I
> figured including NArray might be a reasonable step towards playing
> well with others.
>
>> However, I think the real pushback is that the standard ruby
>> distro is getting rather large and adding something with the limited
>> appeal of narray may not be such a good idea in the effort to keep Ruby
>> trim.
>
> A good point.  Given the limited responses to this thread, I'll take
> it as confirmation of limited appeal.

i personally think this is because no one understands how to use it...  once
you do it's quite useful for many programming tasks...  google
'multi-dimensional array' and you'll see what i mean.  the fact that narray's
give you methods like

   na.shape
   na.rank

alone is a reason to use them instead of some loopy array ctor that hands you
back an array of who knows what dimensionality.

>> probably help, although I am using a version of narray that I have modified
>> so it doesn't do me much good.
>
> Another good point.  Some of the beauty of NArray is the ease of extending
> it, which would again require compilation.
>
> *sigh*  Great.  Now all I've done is publicly concede to arguments against
> including NArray. That's not what I was trying to do...  :)
>
> I guess I was just envisioning ruby grwoing in appeal for large data
> analysis (out of the box).  An alternative to IDL or MATLAB with legs to
> easily do web services and text munging.  If ruby had something like NArray
> built in, one can get a good chunk quickly moving before reaching for
> external packages.  Plus, it'd at least strongly suggest an "approved"
> numerical extension to further build on - perhaps consolidating efforts.

my take on it is this - if you are serious about doing numerical analysis you
have a compiler.  all the guys in my building that don't even understand what
'object' means and who don't see anything wrong with common blocks - they
still know how to compile.  that's because they code in fortran and c.  they
understand how to compile to a larger degree than they understand 'data
encapsulation' or 'abstraction'.  in summary, anyone running on a box that has
no compiler will be thwarted at every step installing numerical applications -
not only ruby.  also, the one-click-installer includes tons of compiled
extensions so i don't see why narray should be any different.

regards.

-a
-- 
===============================================================================
| email :: ara [dot] t [dot] howard [at] noaa [dot] gov
| phone :: 303.497.6469
| anything that contradicts experience and logic should be abandoned.
| -- h.h. the 14th dalai lama
===============================================================================