Re: Anonymous downloads for NCL/NCARG, PyNIO and PyNGL?

From: Charlie Sharpsteen <source_at_nyahnyahspammersnyahnyah>
Date: Tue Oct 12 2010 - 16:35:56 MDT

---------- Forwarded message ----------
From: Charlie Sharpsteen <chuck@sharpsteen.net>
Date: Tue, Oct 12, 2010 at 3:33 PM
Subject: Re: Anonymous downloads for NCL/NCARG, PyNIO and PyNGL?
To: Mary Haley <haley@ucar.edu>

On Tue, Oct 12, 2010 at 2:22 PM, Mary Haley <haley@ucar.edu> wrote:
> Hi Charlie,
>
> I agree that this download process is a bit of a pain.
>
> The reason why we have to do it this way is because we have to be able to report
> to NSF (our main funders) why we should be able to continue developing the
> software. Anonymous metrics don't carry nearly as much weight as being able to
> indicate what types of sites (universities, research companies, weather bureaus,
> government sites, etc) are downloading the software, and how many of these
> downloads are actually from unique users. Being able to say that X number of users
> from Y countries is also very valuable to NSF. We have to prove that NCL is having
> an impact on science, and that is one way of doing it.
>
> There are some folks who have provided packaged NCL binaries for LINUX systems,
> and we just quietly look the other way. :-)
>
> --Mary

Hi Mary,

I was afraid this might be the answer.  I find it very unfortunate
that the NSF heavily weights such detailed download statistics as the
collection of such statistics can drive both users and developers away
from the software.

I was moved to start this discussion by a use case that came up in a
software project I am currently working on.  One of my teammates
needed to access GRIB files from Python in order to extract data.  He
tried using the Python bindings to GDAL and PyNIO and preferred PyNIO.
 However, the software we are making has to be delivered to a client
and one of the questions we will have to answer is "What do I need to
do to get this running on my machine?"

If I can say something along the lines of:

  "Simply run `<package manager> install gdal` and you're ready to rock"

Then the client is happy.  However, if my answer is:

   "Well, first you have to go to earthsystemgrid.org, then you have
to register, then you have to..."

Things don't go so smoothly and I spend the next weeks answering
nervious emails from the client asking why we are using open source,
why Python, why isn't it easier, etc.  So the end result is that we
may choose to use GDAL over PyNIO, even though we would much rather
use PyNIO, just because the deployment process will have less steps
for the end user.

As a developer, I face a different dilemma.  I am perfectly willing to
put in the time to make NCL available as a turnkey installation on the
OS X operating system thus potentially increasing the number of active
users.  However, doing so would require setting up a mirror to serve
the source code -- which is no problem -- but this could potentially
decrease the number of users who retrieve it from earthsystemgrid.org.
 Thus I am in the unfortunate position of having to ask the question
"would donating my time and energy to increasing the availability and
usability of this software end up causing more harm than good?"
_______________________________________________
ncl-talk mailing list
List instructions, subscriber options, unsubscribe:
http://mailman.ucar.edu/mailman/listinfo/ncl-talk
Received on Tue Oct 12 16:36:05 2010

This archive was generated by hypermail 2.1.8 : Tue Oct 19 2010 - 14:38:00 MDT