Re: [Koha-devel] Zebra config problem (still 1)

[Adam Dickmeiss]

Tümer Garip wrote:
> Adam wrote in reply to Paul:
>
> >         cwd=/home/paul/koha.dev/head/misc/zebra/unimarc
> >         15:40:34-03/02 zebrasrv(1) [log] record 0 type XML
> >         15:40:34-03/02 zebrasrv(1) [log] 3129 bytes:
> >         <?xml version="1.0" encoding="UTF-8"?>
> >         <collection
>
> xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance";;
> xsi:schemaLocation="http://www.loc.gov/MARC21/slim http://www.l ...
>
>
> > >                 Your root element is collection. Not record. I don't
>
> think melm will match that. Had you used record as root >>
> element - it should do it.
>
> > >                 It's always a good idea to try things out with
> > >                  zebraidx -s update testrec.xml
> > >                 and see what gets matched.. (Look for the Idx: lines).
>
>
>
> Ooops, this is getting more complicated for me.
> MARC21 slim defines the MARCXML as
> <collection><record></record></collection>. The perl MARC module
> produces XML according to this.
The MARC21 schema also allows <record>..</record>.. The collection 
element was only added as a container (XML allows exactly one root).

> When you run zebraidx with a proper MARCXML you get an error saying
> collection.abs could not be found.
> So you are forced to rename marc21.abs as collection.abs by zebraidx or
> by ZOOM in that sense.
>
> To make things more complicated everything seems to work until you try
> to use the recordID feature.
> If your XML file omits the MARC tags below 10 then RecordID works
> everytime. However if you have tags below 10 out of 50k records I got
> 30% with bad-match error.
>
> So I am trying to eliminate whether its KOHA reproduced records doing or
> not. I'll post the outcome on this list.
I hope to get a sample .. So that I can try this out myself.

/ Adam

> Regards,
> Tumer Garip