Comment 19 for bug 1187433

Bill:

1. Yes, all 3 commits need merging.

2. Yes, the upgrade script is missing. Replacing the guts of the metabib ingest methods is straightforward, the problem is deciding what to do about the reingest.

Mike's reingest approach was working for us, but at a glacial pace (it was going to take at least 13 days to reingest everything and our WAL files were growing at a crazy rate because of the churn of metabib.full_rec & control numbers & authority linking etc; fortunately we still had non-munged metabib.*_field_entry.value columns, so switching to metabib.reingest_metabib_field_entries(id) is about 4 to 5 times faster).

Sites that went live on 2.4.0, or upgraded to 2.4.0 and have already reingested their records, will need to recreate the metabib.*_field_entry rows entirely, because the value columns will have been munged; sites that upgrade from 2.3 or earlier should defer the 2.4.0 reingest and go directly to a metabib.reingest_metabib_field_entries() approach.

Perhaps just echoing that advice at the end of the upgrade script would be enough to get 2.4.1 out the door? Although I would heartily recommend doing something at the 2.4.1 time frame to adjust the 2.3-2.4.0 upgrade script to also prevent people from going through a needless & lengthy reingest at the 2.4.0 portion of the upgrade. Maybe update the metabib function bodies accordingly in the 2.3-2.4.0 script?