[GRASSLIST:1102] Re: Looking for a guru...

[Jack Varga]

Bill,

Could you elaborate?  I know what my perceived "downsides" to ArcGIS 
are, but I'm curious what other perceptions are as well.  I'm not sure 
GRASSLIST is the place to have this discussion, except that Grass has 
similar downsides.  For example, my perception of one major drawback of 
ArcGIS 8.x from the former architecture is that extending functionality 
of any ESRI tool now can only be accomplished by an accomplished 
Windows-based OO software engineer.  Very few natural resource managers 
are also accomplished Windows-based (.NET) OO software engineers.  Very 
few OO software engineers are also accomplished managers of natural 
resources.  It is doubtful that colleges of natural resource management 
will now require 20 credit hours of accompanying object oriented 
analysis and design just so their graduates can extend ArcInfo.  AML on 
the other hand, for all its absurdities, could at least be comprehended 
in a couple/few class assignments.

It's my perception that ArcObjects and the 8.x architecture was designed 
not with respect to Natural Resource managers and scientists, the 
foundation of ESRI users, but with respect to engineers that manage 
utility systems.  Hence 8.x's similarities to the SmallWorld object 
model architecture.  Even the object relational model concept behind 
creating geodatabases really only benefits utilities who need to manage 
n-ary relationships between linear features, yet adds so much complexity.

The lesson here for Grass is perhaps to never lose focus of who the end 
user is and what they are trying to accomplish.  If Grass lacks anything 
at this point it is easy installation and configuration.  The open 
source model, leveraging other open source libraries and initiatives, 
doesn't always lend itself to easy compilation, (i.e., do I build GDAL 
first with libgrass or Grass first and use its libraries to compile 
GDAL?).  What I feel needs to happen (in addition to quicker releases), 
is a Grass installation/compilation project as a wrapper that 
facilitates building Grass successfully and all its constituent 
dependencies, (i.e., libgeotiff, libtiff, gdal, xerces, proj4, etc.) 
much the way the Debian or even Mandrake package managers do, but 
perhaps not tied to specific OS's and/or distributions.

In closing, apologies in advance for the digression.  I am extremely 
indebted to all the current and former members of the Grass development 
team so I hope no offense is taken to any of these comments and only 
wish I could contribute more.

Jack Varga


Bill Dickinson Jr wrote:

> ... though now we are seeing the downside of this version of ArcGIS.