[Proj] Finally: geodesic-1.0 is available

[Gerald I. Evenden]

On Thursday 08 January 2009 12:47:03 pm Frank Warmerdam wrote:
> Gerald I. Evenden wrote:
> > Go to
> >
> > http://members.verizon.net/~gerald.evenden/geod/
> >
> > which puts you into a Doxygen documentation and pointer to distribution
> > tarball at the bottom of the main page.
> >
> > It is ALL there in glorious color.
> >
> > Comments, pro, con or indifferent would be greatly appreciated.  After
> > all that damned effort I'd like somebody to look at it.  ;-)
>
> Gerald,
>
> I did a quick look with the idea of packaging it for OSGeo4W.
>
> Normally distributed source tar files capture everything within a single
> directory, conventionally named the same as the base of the tar file name.
> In your case the tar file has two base directories, html and src.  Perhaps
> you could capture things one level higher?

I will not argue here but I was rather quickly assembling things and as it 
goes I cannot a save out of my development directory because too much other 
non-distributable junk would be assembled.  ;-)  So I have to create a whole 
new temporary file structure to make the tar.

> I would encourage adding a prominent link into the tutorial in Related
> Pages on the main index.html as it is otherwise a bit obscure.

I agree.  I would like to figure out how to manipulate Doxygen to make a new 
tab at the header labeled "User Manual" or "Tutorial".  Dunno if it can be 
done but it is not clear in the rather sparse documentation I have.  I had to 
do a lot of Googling to find examples to fill in the gaps.  :-(  Well 
organized for the developer but needs a little more to help with general 
public documentation.

Also, Doxygen organizes the material inconsistantly at time and I have not 
figured out how to control than.  What I have typed in is all there but not 
consistantly located.  Sigh.

> On my 64 bit linux system under valgrind I was getting messages like:
>
> geodesic computer v. .00
> Free software under X11 license
> geod: earth ellps=WGS84
> geod: p1 77d27'27"W 38d56'30"N
> ==7794== Use of uninitialised value of size 8
> ==7794==    at 0x406A9E: geod_dms2rad (geod_dms2rad.c:73)
> ==7794==    by 0x402014: node (node.c:101)
> ==7794==    by 0x401791: do_look_up (geodesic.c:160)
> ==7794==    by 0x401886: do_commands (geodesic.c:184)
> ==7794==    by 0x40194D: main (geodesic.c:211)
> ==7794==

Me neither.  I ran your sample input and everything functioned normally.  The 
line numbered in dms2rad is blank but even the preceeding line makes no 
sense.

> I stepped through the code and it wasn't obvious why this was occuring.
> When running optimized it was actually crashing in some cases.

So far, I have not had a crash with it for some time.

> I would encourage you to install valgrind on your development linux
> system and use it.  It's a great tool for finding subtle issues you
> aren't even aware of.
>
>    http://www.valgrind.org/

I will try it out.

> I find the mini command language within the program (as opposed to
> a complex set of commandline options) an odd approach.

I do not find it too odd.  For example, I use gp to compute the constants and 
what all and it is often used in the same manner.  Even maxima.  The 
difference is that I do not evaluate math or complex expressions.  In the 
scope of geodesic's operations there is not any need to.  It is just designed 
to allow the user to quickly compute distances or project location on a 
vector and create interpolated geodesic curves or arc.  And as long as the 
user knows a tad about the geographic locations and what a geodesic is then I 
think the internal documentation makes it a complete unit.

> Building was very simple, and I'm interested in preparing a VC++
> makefile to contribute back, as well as preparing win32 binaries for
> the OSGeo4W environment (and I suppose standalone ones).
>
> Best regards,

Thanks for your comments.

-- 
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