> Would there be any objection to overriding the compiled in configuration
> file directory with an environment variable.
Yes. This is compiled in for security purposes. In a perfect world, we would
not have the SVGAlib target. However we do have such a target and maybe
there are other reasons to make LibGGI Programs SUID.
Allowing an environment setting to influence the loading of libraries (which
would be the result of the proposed change) would break system security for
systems that install suid LibGGI binaries.
> This would allow installation in a directory structure other than the
> build structure.
For this reason, we have the patchlib program that will binary-patch the
libraries. When you can write the libs, system security is in your hands
anyway ...
> Example: I configure and build to /e/local/user ( actually
> e:/usr/local on Win98 ). However, I want to install it later to
> /usr/local. Unfortunately, /e/usr/local/etc/ggi is build into the
> executable, and will not run properly on another system W/O
> e:/usr/local) I'd like to override this with an environment variable.
*grin* for that case system security is of course a moot point :-))).
I'd still suggest to just use patchlib. The variable that holds the value is
designed for the patch, so it shouldn't be a problem, execpt maybe a future
version of Win actually checking the EXE checksum in the DLL.
CU, ANdy
--
= Andreas Beck | Email : <[EMAIL PROTECTED]> =
