[Koha] Fwd: Numerically sorting a new Zebra Index

[Senthil Anand]

---------- Forwarded message ----------
From: Senthil Anand <ssenthilanand at gmail.com>
Date: 16 December 2010 19:51
Subject: Numerically sorting a new Zebra Index
To: koha-devel at lists.koha-community.org


Hello,

Running Koha 3.06 on Centos 5.5.

I want to add a new item attribute, canumber to the Zebra Index. It
should be visible on the staff client. All values are numbers ranging
from 1 to around 15000000.

I have done the following steps:

1. In bib1.att

att 8032 canumber

2. In the marc21 record.abs

melm 952$z    canumber, canumber:n, item

3. In ccl.properties

canumber 1=8032

Zebra has been reindexed.

I want to use the canumber in place of acq-date in the Bulk label
barcode generation.

In label-item-search.pl under intranet/cgi-bin/labels . I am
performing a search using the following limits :

"canumber,ge="<textbox data> and

"canumber,le="<textbox data>.

The search actually works and returns results but the results seem to
be sorted as text. ie searching beween 1000 and 1010 returns items
with canumber 1000000 etc. As a test I changed the attribute of
itemnotes in items table from varchar to int(11) in a test machine and
reran the index. Still not getting proper numeric sorting.

The Zebra docs mention that lexicographic sorting is default and
numeric sorting has to be enabled. As I understood it is defined in
the record.abs. How can I change the mappings in record.abs to get
numerically sorted results ? Is the query I am using in
label-item-search.pl correct ?

Regards,
Senthil