[Proj] New release of libproj4

[Gerald Evenden]

In the testing of the PEI data I could demonstrate that libproj4 could 
do
micron round-trip (or better) but I am usually faced with the problem 
that
the test data I get is only to 10mm.  In this case, the xy was 1mm and 
the
geographic to 0.0001".  Right on the edge of demonstrating precise
computation.

Unfortunately, as Snyder pointed out to me one time (and I agree),
round trip does not prove the correctness of the code nor algorithm.
However, I do use it as an indicator that I am on the right track when
developing a projection.

As an aside, when using lproj to prove the round-trip ability of a
projection, pipe the forward output to the inverse in binary so as to
avoid formatting issues.  For example:

lproj $CTL test.dat -o |  lproj $CTL -I -i -W8

where test.dat contains geographic points.  Options -o and -i do
binary output and input respectively and -W8 gives 8 fractional
digits to the seconds of output.  $CTL contain projection +opt control.

On Feb 5, 2005, at 12:23 PM, Clifford J Mugnier wrote:

> Gerald,
>
> Round-trip precision of 0.1 mm is considered standard for geodetic
> transformations.  (X,Y -> Lat,Long -> X,Y)
>
> 19th Century precision was considered standard (at the time) at 0.1 
> meters.
> That's why a lot of old projection equations need an iterative loop 
> for the
> inverse nowadays.
>
> Cliff Mugnier
> LSU
_____________________________________
Jerry and the low riders: Daisy Mae and Joshua.
"The whole religious complexion of the modern world is due to the
absence from Jerusalem of a lunatic asylum." Havelock Ellis, 1914