[Proj] Why?

[Gerald I. Evenden]

On Monday 27 October 2008 11:30:16 am Frank Warmerdam wrote:
> Gerald I. Evenden wrote:
> > On Monday 27 October 2008 9:48:36 am support.mn at elisanet.fi wrote:
> >> "Gerald I. Evenden" <geraldi.evenden at gmail.com> kirjoitti:
> >>> One burning question has come up in my mind about the need for inverse
> >>> solutions for non-grid-system projections like Aitoff, Hammer, etc.:
> >>> why?
> >>
> >> For our part we need the inverse for every projection, since the users
> >> want to get their coordinates and other information just by clicking on
> >> the maps. For us those projections without forward and inveres are not
> >> usable at all. We currently simply discard all projections which do not
> >> have inverses. So we need them both all the time. And of course in the
> >> view point of having a general projection package there could be all the
> >> inverses also defined as far as it is practical.
> >>
> >> Regards: Janne.
> >
> > I cannot say that I understand your practices but outside of cadastral
> > operations I do not see the usefulness of more than a four or five
> > projections---even less for screen display.
> >
> > Really, for point and click operations one only needs equidistant
> > cylindrical and polar.
> >
> > For publication use anything that suits the purpose because you do not
> > need to retrieve data from such output.
> >
> > As I said before, I do not comprehend the why of the "needs" of your
> > users and/or the item you are selling.
>
> Gerald,
>
> I'm with Janne on this.  In typical GIS applications (I am often dealing
> with users of UMN MapServer) it is necessary for the projections to be
> invertable in order to use them systematically.   From a practical point of
> view there is little point in asking why folks use a particular projection
> in a particular way.  If the projection exists it will get used in a
> variety of ways - some of which are objectively somewhat silly.
>
> But to be well behaved for applications built on it, it is very helpful
> for the library to support inverses for all projections that might get
> used.
>
> Best regards,

Let's go back to basics: I have not restricted or deminished the use of 
libproj4 from the level of its current distribution.  It is still free and 
very open software and does everything that the old version does.

The only problem is that if someone wants to use some new features they may 
run into another license such as GPL.  To the best of my interpretation, 
using the GPL material does not cost anything and has no restriction on 
personal usage.  Again, if I am wrong in my interpretation of GPL please 
correct me.

Remember, there hase been a current gsl/GPL requirement with libproj4 if you 
want to use all the projections in the library.  This does not seem to have 
been the end of the World.

Where I suspect we are getting into conflict is with commercial use of 
libproj4 and the monkey wrench that GPL introduces into the sale of a product 
that may contain material with GPL.  To answer this problem I suggest that 
there are probably many suppliers of similar libraries which will do the same 
thing and are readily patchable as a replacement.  This may add some license 
fee expense  to the product but, alas, that is a business problem.

Lastly, to reemphasize my reason for using gsl is that the gsl procedures 
involved do a better job and are more robust than simple application of 
Newton-Raphson.  Secondly, there is the ability to find inverses of 
projections where practical determination of derivatives is impossible.

-- 
The whole religious complexion of the modern world is due
to the absence from Jerusalem of a lunatic asylum.
-- Havelock Ellis (1859-1939) British psychologist