hi again,
after taking a closer look, I rejected my original tree-searching idea. It is probably not worth the fuss, because it cannot be done well with any simple kind of path-walking performed during loading of a library. If my (hazy) idea was to be rendered into a real design, there would need to be an extra knowledge used -- gathered beforehand. This preprocessing is needed both to speed up the search, somewhat like locatedb or ls-R of texhash, and to filter out obsolete versions, resolve name clashes, etc.
So if there was to be a kind of personalized pdb on every Pd user's hard disk, why not to design a standard mechanism for registering libraries with a version number, and a full list of external classes: their names, help file locations, `assistance' descriptions, links to reference pages, etc.
This user-pdb or system-pdb could be updated during library installation pass, and then consulted by the sys_load_lib() call in Pd -- before or after or instead of searching .../pd/extra and the -paths.
The next step would be to connect user-pdb to graz-pdb (or mirrors) ala deb :)
|
Krzysztof Czaja wrote: ...
flat directory. I would rather put my stuff somewhere down the tree.
...
- Design an optional tree-searching (or full-blown globbing)
mechanism for open_via_path() to be used whenever loading libraries (and possibly abstractions) is involved.