[OpenAFS] Small suggestion for RH spec file?

[Derrick J Brashear]

On 29 Dec 2002, Nathan Neulinger wrote:

> Seems like having a separate RPM for cellservdb would be easiest. 
> 
> Then you can have TWO rpms for the 2 cases that cover 99% of the users:
> 
> 1. Local only - this RPM would have no CellServDB.global
> 2. Full participant - this RPM would have a full CellServDB.global

Hmmm. This might actually be the solution, though there is one problem:
how do you bootstrap? I don't think rpm has a way to depend on "a or b". I
guess you could make both rpms "provide" something, but they also need to
conflict.