LOADPATH and cluster.pan.includes

[wpoely86]

When using ant, you have pan.loadpath and cluster.pan.includes. After reading the documentation, it still not very clear to me what these two do and how they can be combined? If I understood it correctly from @stdweird, you can 'prepend' search locations of cluster.pan.includes with pan.loadpath?

It would be nice if somebody could clarify the documentation a bit.

[stdweird]

@wpoely86 they are combined in the following code where the filesystem lookup of the pan templates happens:
https://github.com/quattor/pan/blob/master/panc/src/main/java/org/quattor/pan/repository/FileSystemSourceRepositoryImpl.java#L63

(and empty string is always part of loadpath)

includedirs are set outside pan, loadpath inside pan (except for the ant variable, which has no equivalent on panc commandline)

also with loadpath, you cannot break out of the includedirs.

when includedirs are not orthogonal to the loadpaths, you can expect strange issues (but they are not bugs; the behaviour is 100% predictable)

what would you like to see clarified? you want an example how it works (and how it can go wrong? 😉

[wpoely86]

Some examples/guidelines how to (ab)use it would be nice.

For example, the empty string in loadpath: is it first or last or something else? I assume last?