[OpenAFS] a familiar problem compiling aklog?

Russ Allbery rra@stanford.edu
Tue, 13 Dec 2005 20:42:05 -0800

edwin <edwin@itasca.net> writes:

> Ken has suggested to me that changing include orders might fix the
> com_err problem with aklog.  I'm willing to futz with that, but it seems
> that there must be a known solution somewhere.  I don't think my systems
> are that non-standard, so I'd expect a lot of people to be bumping into
> this problem.  If they aren't then

> 1) people aren't running AFS on Linux but only on Solaris, etc.
> 2) people aren't compiling AFS on Linux from source but installing 
> binaries (which is a bit of a crock--somebody's compiling the binaries 
> initially somewhere, and they'd have the same problem)
> 3) people aren't using the openAFS 1.4.0 source but some earlier version 
> which does not have the problem
> 4) openAFS installed base is small enough (in terms of deployed
> instances, not end-users) that just aren't that many people who've been
> down this road before.

5) People already have an aklog and are still using that one rather than
switching to the one that ships with OpenAFS yet.

I'm going to start dealing with these issues with 1.4.1 when asetkey is in
OpenAFS as well.

> OpenAFS is in a pretty small group of programs I've encountered that did
> not ./configure && make && make install without tweaking.

This does normally just work.  aklog was *just* added to the tree in the
1.4.0 release and we're still working out the kinks with trying to link to
both Kerberos and AFS libraries at the same time.

Russ Allbery (rra@stanford.edu)             <http://www.eyrie.org/~eagle/>