From: Demian K. <dem...@vi...> - 2013-04-04 15:35:12
|
> At the moment I do not need change tracking but just in case I want a > complete compatibility layer in the future: where can I find info about > the necessary database writes? You might find the ChangeTracker class in the VuFind source helpful: http://vufind.git.sourceforge.net/git/gitweb.cgi?p=vufind/vufind;a=blob;f=module/VuFind/src/VuFind/Db/Table/ChangeTracker.php;h=bb5e428655cad1e50b40ef3b8042ffe9ca629140;hb=HEAD The markDeleted() and index() routines show how the table is populated. Note that there are some outstanding issues related to time zones; presently, the code may misbehave if you access the same database from servers in different time zones, etc. > I guess this uses JSON? I think MARC, as well as my own data, has some > characters that are illegal in XML. Or how does SolrMarc get the > original MARC data into the index (over HTTP)? SolrMarc escapes the three MARC characters that are not legal in XML, and then VuFind unescapes the resulting data. > That is very good news since we only have one hosted server. But are > there no database and/or Solr schema changes between the major Vufind > versions? In particular (since you answered it for the index) can I use > the same MySQL database for both versions, in parallel? There are some differences between VF1 and VF2, but everything unique to VF2 is an addition rather than a change. You need to upgrade your 1.0 database to 2.0 in order to run both versions in parallel, but as long as you avoid a couple of steps (don't convert database encoding to UTF-8, and don't start encrypting user passwords and ILS credentials) both versions will work on the same database. As soon as you are done with 1.x, then you can go ahead and re-run the upgrade script to improve security and character representation. - Demian |