Re: [htdig] irrelevant pages in search

Subject: Re: [htdig] irrelevant pages in search
From: Gilles Detillieux (
Date: Tue Nov 23 1999 - 11:57:03 PST

According to Hartmut Steffin:
> i had a bad database. i used the - more or less - automated unchanged script
> "rundig". if you do something autotomatic, look at the messages! htmerge
> failed (with the simple message "htmerge failed". you should do something
> about this!) due to unsufficient space in /tmp while in the part the
> standard command "sort" does its work. despite the man page, sort and/or
> htmerge does not seem to honour the TMPDIR variable which IS properly set
> and exported in "rundig" and points to a hd with sufficient space.

When the sort fails, htmerge should end with "htmerge: Word sort failed"
as the error message. Preceeding this, there should be a more descriptive
error message from the sort program explaining the reason for the failure.
If there isn't, either your system's sort program is at fault here, or the
problem is something other than sort running out of disk space.

The way htmerge honours the TMPDIR environment variable is it passes that
value to sort using the -T option. If your system's sort program doesn't
honour that option, and doesn't use TMPDIR itself either, then you may want
to install GNU sort in its place, or modify htmerge/ to use another
option if your sort program supports some other option. In the case of BSD
systems, the sort program actually uses -T for something else, so you can't
define TMPDIR on these system due to this conflict.

> i did not have the time yet to hunt the problem down to the specific
> program. i just wanted to let you know.
> in ANY case,
> 1. htmerge should do a better error message (I even used -v)

We're open to suggestions, but if the problem is the sort program that fails
silently, there isn't much that htmerge can do to guess at why.

> 2. htsearch should be able to identify a corrupt db

I too would like to see more error checking to detect such problems, but
I wouldn't know where to begin in adding code, and what to look for in terms
of database problems. Anyone else have any ideas?

Gilles R. Detillieux              E-mail: <>
Spinal Cord Research Centre       WWW:
Dept. Physiology, U. of Manitoba  Phone:  (204)789-3766
Winnipeg, MB  R3E 3J7  (Canada)   Fax:    (204)789-3930

------------------------------------ To unsubscribe from the htdig mailing list, send a message to You'll receive a message confirming the unsubscription.

This archive was generated by hypermail 2b25 : Tue Nov 23 1999 - 12:08:58 PST