[GRASS5] [bug #1069] (grass) Re: GRASS and Idrisi file compatibilities

[Request Tracker]

this bug's URL: http://intevation.de/rt/webrt?serial_num=1069
-------------------------------------------------------------------------

Hello Eric,

Many thanks for your comments, recently sent to david.a.hastings at noaa.gov.
I currently find that I can't compose replies in that email system from
here (UN-ESCAP,Bangkok, where I currently work), so am replying from here.

Since I created a dual-boot Linux-Windows system in 1995, I was able to
work with the same uncompressed 8-bit cell files in Idrisi and GRASS, thus
de-facto file-level interoperability.  Of course, the headers were
different, but these could be kept in separate directories, so no problem
there.

(If you're familiar with the origins of the Open GIS consortium, and its
avoidance of GRASS, the existing interoperability of GRASS with another GIS
is somewhat interesting.  Because of that history, it would be great to see
de-facto GIS interoperability demonstrable more strongly from GRASS, if
that were not much work.)

It would have been nice not to be forced to have automatic compression of
GRASS cell files when they were created.  However, this is only a minor
inconvenience to uncompress a cell file to make it compatible with Idrisi's
uncompressed format.  (Idrisi also has run-length encoded file compression,
but this used to be error-prone, so I don't know anyone that compresses his
Idrisi files.)

The next hurdle would be integers.  To me, the GRASS integer file format is
1980s concepts of portability run wild.  Or, perhaps, not updated with the
times.  Many current programs handle either byte ordering for integers.  If
this were allowed in GRASS, both PC and UNIX byte ordering could be handled
by the software, rather than the (easy, but very bothersome) playing with
dd....conv=swab.  Also, as so much integer data is merely 2-byte integers,
a simple 16-bit integer format in GRASS would have been helpful - again
allowing for either byte ordering.  (I'd also let the 16 bits be signed or
unsigned with another header flag...)

If GRASS had a 2-byte integer format that could handle either byte ordering
plus signed or unsigned integers, its file-level portability would markedly
increase.

For larger integers, GRASS might keep its current format, which is a
non-standard enhancement useful when numbers get too large for 16 bits.

Perhaps the optional header lines for integers could be something like:

intsigned
intunsigned
motsigned
motunsigned

with int and mot used instead of pc (or intel) and unix (or motorola) to
avoid trademark issues, as well as to avoid the confusion of the various
unix platforms (e.g. alpha)..

Perhaps such terms (or an equivalent) can go into the "format" field
instead of the adjusted "number of bytes per pixel - 1" normally used.

Floating point data might not be worth worrying about yet.  Idrisi actually
started with 6-byte floats, based on Turbo Pascal (I believe that was their
original design platform).  Later they went with 4 and 8 byte floats.

I hope this isn't too "windy."  But I think it better to lay out my ideas
for you to eliminate, than for me to be too brief and have you guessing the
details.

With best regards,

David Hastings
Geographic Information Systems
Space Technology Applications
UN Economic and Social Commission for Asia and the Pacific
UN Building
Rajademnern Nok Avenue
Bangkok 10200, THAILAND

email: hastingsd at un.org
tel: +66-2-288-1457


--- Headers Follow ---

>From hastingsd at un.org  Thu May 30 16:48:15 2002
Return-Path: <hastingsd at un.org>
Delivered-To: grass-bugs at lists.intevation.de
Received: from mail.intevation.de (aktaia [212.95.126.10])
	by lists.intevation.de (Postfix) with ESMTP id 8BDB413965
	for <grass-bugs at lists.intevation.de>; Thu, 30 May 2002 16:48:15 +0200 (CEST)
Received: from mx1.un.org (mx1.un.org [157.150.197.100])
	by mail.intevation.de (Postfix) with ESMTP id 1D5E61B71A
	for <grass-bugs at intevation.de>; Thu, 30 May 2002 16:48:14 +0200 (CEST)
Received: from un-mailhub-01.un.org (un-mailhub-01.un.org [157.150.197.17])
	by mx1.un.org (8.11.0/8.11.0) with ESMTP id g4UEpAa28556
	for <grass-bugs at intevation.de>; Thu, 30 May 2002 10:51:10 -0400 (EDT)
Subject: Re: GRASS and Idrisi file compatibilities
To: grass-bugs at intevation.de
X-Mailer: Lotus Notes Release 5.0.3 (Intl) 21 March 2000
Message-ID: <OFCCC34CDD.3BAB25F8-ON47256BC9.0031D261 at un.org>
From: "David A Hastings" <hastingsd at un.org>
Date: Thu, 30 May 2002 16:51:40 +0700
X-MIMETrack: Serialize by Router on un-mailhub-01/UNO(Release 5.0.8 |June 18, 2001) at
 05/30/2002 10:48:11 AM
MIME-Version: 1.0
Content-type: text/plain; charset=us-ascii
X-Spam-Status: No, hits=0.0 required=5.0 tests= version=2.20
X-Spam-Level: 

-------------------------------------------- Managed by Request Tracker