[xml-cppdom-devel] Versioned directories on Windows

[Patrick H. <pa...@13...>]

Sorry for the cross posting, but this involves both CppDOM and GMTL. I
recently made the statements below on the vrjuggler-devel mailing list an=
d
have received no comments, criticisms, or feedback. This is an important
issue that needs to be resolved, so here is what I said:

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=


Versioned Header and Data Directories on Windows
------------------------------------------------

Getting back to the parallel installation topic, Windows has been updated=
 to
created versioned DLLs and to generate .fpc files for building against VR=

Juggler from the command line (yes, that is finally supported again).
However, the header and data directories are not currently being versione=
d
the way that they are on non-Windows platforms, and a decision needs to b=
e
made regarding whether it is necessary to do that.

My vote on this topic is not to version the header or data directories on=

Windows. Deployment on Windows is fundamentally different than on other
operating systems because there is no equivalent of /usr/include (or
/usr/local/include) or /usr/share (or /usr/local/share). Rather, software=

on Windows usually goes into C:\Program Files\<directory> where <director=
y>
can be named with a version. This makes parallel installations easier sin=
ce
the files are separated at a higher level in the directory structure. It
also makes it easier to handle Visual C++ project files since they don't
have to be updated to track the name of the include directories (which co=
uld
require a lot of work since there are would be one /I option needed for e=
ach
Juggler module). Structuring things this way is what the Boost Consulting=
's
slick installer for Boost on Windows does, and I would apply this same
argument to CppDOM and GMTL.

=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=
=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=3D=


Thoughts?

 -Patrick


--=20
Patrick L. Hartling                    | VP Engineering, Infiscape Corp.
PGP: http://tinyurl.com/2oum9          | http://www.infiscape.com/