[Proj] minor proj bugs
warmerdam at pobox.com
Wed Jul 28 19:31:46 EDT 2004
Britton Kerin wrote:
> Not sure this list is alive (the archive link on the web page doesn't
> work). The web interface to bugzilla isn't working either though, so here
Have have removed the old archive link and noted that the subscribe/unsub
link also leads to archives.
> 1. There is a discrepency between the pj_geocentric_to_geodetic
> and pj_geodetic_to_geocentric prototypes in proj_api.h and the
> corresponding definitions in pj_transform.c. The second argument in
> the signatures in the definitions is 'es', which I assume stands for
> eccentricity squared. The first thing the functions do is compute 'b'
> (I assuem the minor axis treating es as eccentricity squared, so I
> guess eccentricity squared s what it is. The prototypes in proj_api.h
> have a second argument called 'ra' however, which is confusing. It
> looks to me like the names in proj_api.h should be changed from 'ra' to
Agreed. I have fixed this in CVS.
> 2. The description of the -format option in the cs2cs man page contains an
> unecesary sentence:
> Format is a printf format string to con-
> trol the form of the output values. For
> inverse projections, the output will be
> in degrees when this option is employed.
> If a format is specified for inverse pro-
> jection the output data will be in deci-
> mal degrees. The default format is
> "%.2f" for forward projection and DMS for
Fixed in CVS.
> 3. Advertise the new general transformation capabilities (cs2cs and
> corresponding capabilities of 'projPJ' type in library) better! They
> are really cool and I almost missed them entirely.
Advertise where? I can't update the postscript/pdf manuals as they are "readonly"
at this point. But the new stuff is listed in the pj_init man page and cs2cs
has a SEE ALSO'ed man page.
> 4. Some of the API routines look redundant with each other now. For
> example, if I understand correctly: pj_datum_transform is now subsumed
> into pj_transform? And pj_fwd and pj_inv are also simple versions of
> pj_transform? This redundance should probably be discussed somewhere
> to prevent confusion. Or, if they are different, the differences
> should be described.
pj_transform() is built by using the various component transformations. The
pj_datum_transform() isn't really intended to be too public, and I don't
think I have documented it anywhere.
> proj is excellent in general though, definately worth keeping it up to
> speed and polishing up the docs.
I set the clouds in motion - turn up | Frank Warmerdam, warmerdam at pobox.com
light and sound - activate the windows | http://pobox.com/~warmerdam
and watch the world go round - Rush | Geospatial Programmer for Rent
More information about the Proj