[OpenAFS-devel] [Fwd: History of AFS]

[Ted Anderson]

I can't comment much on the current state of the code.  I do know that a
big problem with maintainability was due to the #ifdefs used to handle
the wide diversity of ports.  This caused a constant pressure to
"mess-up" the code and there was little time available for cleanup.

As Derrick mentions, the deeper origins of the current AFS code base go
back to the AFS2/AFS3 development.  This is approximately where the
AFS/Coda split occurred because ITC/CS took different paths at this
point.  The ITC wrote a kernel resident cache manager for AFS3 while the
CS department continued with the Vice/Venus architecture which used only
a few small kernel hooks.  A lot of the ITC cache manager code remains
in the OpenAFS source.  The server side code probably has deeper roots,
but the vlserver, kaserver and ptserver are "new" in in AFS.  I don't
know the origins of the file volume server split.

The point of all this is that you need to go back before mid-1988 when I
arrived on the scene, as the kernel cache manager rewrite, Ubik
development and the separate vlserver were mostly in place at that time.
To find out first hand about this you'll need to talk to Mike Kazar,
Vasilis, Dave Nichols, John Howard and some of those folks.

Ted Anderson