Re: [htdig] question about rundig.3.2.sh

[Gilles D. <gr...@sc...>]

Earlier I wrote...
> According to Greg Fenton:
...
> > Now, nowhere else in the script is anything done with the .index file.
> > 
> > So, does this line need to be uncommented?
> > Should this be a "cp" instead of a "mv"?
> 
> As the comment says, since "these are only used by htdig for update digs
> and we always use -a, we just leave them as .work".  So, unless you plan
> to run htdig without -a, you shouldn't need the non-.work version.  At
> least, according to the attrs.html documentation for this snapshot, the
> doc_index attribute is used only by htdig.
> 
> However, in looking at the source code, I do see a discrepancy there.
> It seems there are references to doc_index all over the place, in
> htdig, htload, htmerge, and htpurge (which all make sense to me), but
> also in htdump, htstat and htsearch (which doesn't make sense to me).
> 
> Geoff, would you care to comment?  Why do these programs which purportedly
> don't need db.docs.index, or at least shouldn't need it, still seem to
> require it.  Is there a problem with the way the DB handling code is
> structured right now that requires us to open the index even if we don't
> use it?

OK, I've confirmed by running htsearch that this is indeed a problem.
If you run htsearch without moving db.docs.index.work to db.docs.index,
or copying it, htsearch complains:

Unable to read document index file '/var/lib/htdig/db.docs.index'
Did you run htdig?

So, htsearch won't run without access to a file it ought not to need, and
which the documentation claims it doesn't use.  Is there an easy fix to this?

-- 
Gilles R. Detillieux              E-mail: <gr...@sc...>
Spinal Cord Research Centre       WWW:    http://www.scrc.umanitoba.ca/
Dept. Physiology, U. of Manitoba  Winnipeg, MB  R3E 3J7  (Canada)