kafs.h from heimdal or from arla

[Rasmus Kaj]

Am I the only one who has the problem that #include <kafs.h> in the
arla sources gets a kafs.h that was installed by heimdal rather than
by arla, and which lacks some symbols (the first that the complier
complains about is VIOCNOP).

I can think of three obvious solutions for this:

 1) Add a file kafs-arla.h which is included right after kafs.h and
    checks if some definitions are missing and then adds them.
  + Relatively simple, especially since the missing things is macros.
  - Rather ugly.

 2) Update the kafs.h in heimdal to include everything that arla
    needs.
  + Simple.
  - Long turn-around time, since it not a solution before the next
    release of heimdal.  What about next time?

 3) Remove kafs.h and the entire libkafs from heimdal.  This requires
    some kind of plugin architecture, so kauth can still create afs
    keys for the user.
  + Theoretically pleasing.
  - Requires quite a lot of work.

So, what do you think?  There is certainly other pros and cons to each
possibility, and probably other possibilitys as well ...


-- 
Rasmus Kaj ----------------------------------------------- rasmus at kaj.se
 \                   An NT server can be run by an idiot, and usually is
  \------------------------------------- http://www.stacken.kth.se/~kaj/