Problems rebuilding history

[Alex Kiernan]

Russ Allbery <rra at stanford.edu> writes:

> Jeffrey M Vinocur <jeff at litech.org> writes:
> 
> > Russ, I'm forced to conclude there must be a bug in the new overview 
> > implementation :-/
> 
> > This is CURRENT from Oct 1.
> 
> > After doing the following:
> 
> > # rm -rf spool/overview/* db/history*
> > # bin/makehistory -b -O -e -l 100000 -F
> > # bin/makedbz -s 750000
> > # mv db/history.n.dir db/history.dir
> > # mv db/history.n.hash db/history.hash
> > # mv db/history.n.index db/history.index
> > # chown -R news.news db/ spool/overview
> 
> I'm not sure how the new overview method could cause problems with
> generation of the history file; the code paths are pretty distinct.

AFIACR they're completely disjoint, in fact I've just remembered that
makedbz hasn't gotten moved to the history API at all yet so its
almost completely stand alone.

-- 
Alex Kiernan, Principal Engineer, Development, THUS plc