I came up with a new idea of a small enhancement - right now all
load/save/... operations require a file name which must be accessible
exactly as specified. This makes using Clips embedded within a program
hard, which may be relocatable and where I don't know where is is
started from.
If my Clips code says somewhere (load bla.clp) then bla.clp must exist
in the same directory as the current dir of the executable - which may
be correct or maybe not.
Alternatively, I can specify (load /some/path/to/my/clp/files/bla.clp)
but I don't want to hardcode this in my CLP main file.
Solution would be to have something like a search path - there could
be an env var CLIPSPATH which accepts a PATH in UX or Windows
style like /a/b/c:/d/e/f:/some/path/to/my/clp/files which will be used
in all load/save/... functions.
This way my executable can set this path appropriately (depending on
where it is installed) and in the CLP code I still can say (load
bla.clp) which then looks through the SearchPath if there is one.
Seems fairls easy to do, no side-effects, existing progs should not be
affected.
sysdep.c patch
Generally works with g++/Linux and mingw or MSVC on Windows.
Slightly modified the behavior in a way, that this only applies to files which are to be opened for reading as I probably don't want to create/write files at a location which may be variable.
Meanwhile I still have the code in a locally modified version but don't use it anymore because of a different approach of loading CLP files. However, it still might be interesting to other people ...