All, At Chris's suggestion, I'm moving this to the development list. It seems to me that Henri-Damien is correct in that exact would be better for the sn and bc indexes but that the phrase might work better for call numbers. Since I am very much a newbie at this, I'd like to hear from those more experienced about the best way to proceed. Do I need to put this discussion on bugzilla or just the outcome of the discussion when we're ready to proceed? We are very interested in contributing but don't want to create any problems when trying to fix something. Thanks, Linda On 1:59 PM, LAURENT Henri-Damien wrote:

Le 06/06/2011 20:57, Chris Cormack a écrit :

...

...

On 7 June 2011 06:53, Linda Culberson<lculber@mdah.state.ms.us> wrote:

...

...

The problem, at least in 3.4, does exist, but changing the query from sn,wrdl to sn,phr and bc,wrdl to bc,phr seems to fix it.

But I'm still wondering if there are others - besides call numbers - for which this applies.

I apologize for the multiple emails. I should do more research before I email. I'll try to keep quiet now.

Hi Linda Multiple emails are fine, but perhaps this discussion is better on the koha-devel list. I think its not a new problem (i think 3.0.x up did this) but that doesn't mean its right.

I think phr for barcode, sn, etc makes more sense and I would encourage you to widen the scope of your bug to include them. The fix is simply changes in the template so should be relatively straightforward to implement and would be a welcome fix for many libraries I think.

Chris Well, it all depends on what you search, but ext could be really what you are looking for. With phr you could look for : AAAA/BB And find AAAA/BB C AAAA/BB D or even C AAAA/BB where ext would match the exact subfield... But one default for that is that you cannot search both ext and * afaik (could be quite easy to test if anyone interested. In a yaz-client : f @attr 4=1 @attr 6=3 @attr 5=2 "your callnumber" )

and that ext search requires the completeness set to 1 on the index type you use in default.idx ... Which is commonly the case when you use p... Moreoved, consider that using p should not replace w indexing, but be added. Since there are places in the code or in database settings where lcn can be searched directly without using phr modifier. And if you donot have indexed with word... Then the search will always fail which would be worse than the problem it fixed. Hope that helps. -- Henri-Damien LAURENT

-- Linda Culberson lculber@mdah.state.ms.us Archives and Records Services Division Ms. Dept. of Archives & History P. O. Box 571 Jackson, MS 39205-0571 Telephone: 601/576-6873 Fax: 601/576-6824