https://bugs.koha-community.org/bugzilla3/show_bug.cgi?id=44064 --- Comment #1 from Martin Renvoize (ashimema) <martin.renvoize@openfifth.co.uk> --- Hi Katrin, Marcel and all, A lot has happened since I filed this, so here is where it stands and what I would like to do next. The indexes are no longer part of this bug. Point 5 of the description is done or attached: rcn-volumes (bug 39287), the missing fields in rcn (38252), rcn-components (35330) and Show volumes for serials (42027), plus bug 35117 (a record without a 001) and bug 44063 (the component sort). What is left here is how a host is matched to the records that link to it: get_components_query, get_volumes_query and the links that the stylesheets build. 1. Match the (003)001 form that the specification defines. 2. Decide what a $w without an organization code means, and whether the MARCOrgCode should be assumed. 3. Handle the prefixes some agencies put on their 001 (for example ocm and ocn with OCoLC). 4. Consider the 035 numbers of the host (bug 21472). Records imported from another system often have their original 001 moved to the 035 (for example an (OCoLC) number), while the $w of the linked records still uses that number. Only the 001 and 003 of the host are used today, so those links are not found. 5. Build the query in one place. The Analytics and Show volumes links of the staff interface and OPAC are built in the stylesheets from the 001 only, so they ignore the 003 and can give a different result from the Components tab. 6. Quote the values in get_volumes_query like get_components_query does. I would like to keep the other direction, finding the host of a record from its $w (get_marc_host and the Set and In links, bugs 27507, 29268 and 21472), apart from this bug. They have not moved for a long time and should not hold this one up, so I am removing them as dependencies here and leaving them as see also. A question for Katrin and Marcel, as this is the part the specification does not answer. With a host that has a 001 and a 003, a linked record with a bare $w (no organization code) is only found today if its own 003 is the same as the host's. With a bare $w and no 003 it is not found. I checked this on Zebra, although the Multipart Bibliographics wiki page says it is found. What should happen? a) Keep it as it is: a bare number only matches records of the same agency. b) Also match a bare number on records without a 003, assuming the same agency. This is what the wiki page describes, but it can match the wrong record if the numbers of two agencies collide. c) Only match the (003)001 form, as the specification defines it. I have no strong preference, so I would like your views. In the meantime I will start on 5 and 6, which do not depend on the answer. -- You are receiving this mail because: You are watching all bug changes. You are the assignee for the bug.