[Proj] Some typos in the libproj4 manual

[Clifford J Mugnier]

The statement, "By stating that 2 decimals are more than enough, because we don't do
nothing below the centimeter level" is incorrect.
 
In the United States, for the legally-defined State Plane Coordinate Systems, the controlling document for algorithms and computational precision is NOAA Manual NOS NGS 5 by James E. Stem.  That is the official document.  It can be downloaded in a pdf file from http://www.ngs.noaa.gov/PUBS_LIB/ManualNOSNGS5.pdf  
 
In that document, the requirement is for computational accuracy to the tenth of a millimeter (0.0001 meters).
 
That is not getting carried away, it is complying with a National Specification.
 
Cliff Mugnier
LOUISIANA STATE UNIVERSITY


________________________________

From: proj-bounces at lists.maptools.org on behalf of OvV_HN
Sent: Thu 02-Oct-08 04:09
To: PROJ.4 and general Projections Discussions
Subject: Re: [Proj] Some typos in the libproj4 manual



----- Original Message -----
From: "Gerald I. Evenden" <geraldi.evenden at gmail.com>
Sent: Wednesday, October 01, 2008 10:55 PM
Subject: Re: [Proj] Some typos in the libproj4 manual

....
>> problems in the manual.
....
> I will put the current version on my site by this weekend.  Some
> corrections
> may have escaped due to my occasional  sloth.  My apologies in advance.

Thanks!


>> Another suggestion: give a numerical example with at least 5 decimals
>> plus
>> the exact command line code for each and every function.
> What do you mean by "every function."  In term of precision of functions
> that depends upon what the function is returning.
....
> On occasion, I think some get a little carried away with precision.  ;-)

Working at the nanometer level has absolutely no significance for geodesy.
That part we do agree upon.
But look for instance at the test files in the GNU GSL distribution.
One has to be able to check the correct functioning of all available
functions in all relevant or in all supported regions.

It would be nice to give some real world examples, using parameters of
actually used projections, but exactly at that moment some people will drift
off from the intents of a test example.
How? By stating that 2 decimals are more than enough, because we don't do
nothing below the centimeter level.

What is needed are abstract examples or real world examples, just examples,
with a lot of decimals, so that the user can check whether his compile, his
microprocessor, his porting, his whatever, are done perfectly.

Please allow me to repeat:
One has to be able to check the correct functioning of all available
functions in all relevant or in all supported regions.


Oscar van Vlijmen



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.osgeo.org/pipermail/proj/attachments/20081002/a054d65a/attachment.html>