All my questions were aimed towards a binary distribution.
F.e. our current 1.0.1 for OSX package on sf.net download
page.
[ I know these can be controlled when building from source
and running 'make install', this is exactly how it's working
here locally on OS X, Linux and Windows. ]
For the rest: why would I have to modify an environment
to run a simple executable?
It's very easy to avoid and very unnatural, so why
make it so complicated and not work "out of the box"?
Well, then I'd guess that the rpath should be embedded into the
harbour executable and the executables generated by it. Adding
"--rpath=<your desired path>" (or maybe "-Wl,rpath,<your desired
path>", not tested) to the gcc command line should do the trick and
save you of having to specify LD_RUN_PATH each time.
This is an obscure (for me, at least) feature of gcc, and I could very
well be wrong, and I shiver at the idea of adding the current
directory to the run path. But if this is in the line of what you
want, it could be worth to investigate.
If this is not in the line of what you want, I'm afraid I completely
missed your point :-)
Above is an even worse nightmare :) I'd prefer to not
embed _any_ absolute locations (or any local environment
specific setting whatsoever) anywhere in a program.
An executable always knows the directory it resides in,
and this should be enough for it to find everything needed.
This is also a main rule for all desktop apps in OSX.
Brgds,
Viktor
_______________________________________________
Harbour mailing list
Harbour@harbour-project.org
http://lists.harbour-project.org/mailman/listinfo/harbour
