Update arclight to v1.1.0
corylown opened this issue · comments
Cory Lown commented
- Work in progress PR here: #607
- I don't think we need or want to make any Solr config or schema changes, but we should review what changed in arclight v1.1.0 and decide if any of it is relevant.
-
Coordinate with @laurensorensen to re-index VT. (There are some breaking changes to breadcrumbs and we can remove some patching around the change if we re-index.)will handle as part of #627 -
If thewill handle as part of #627Arclight::Parent
override is removed from #607 (that added the global_id method back to the class to keep breadcrumbs working) then we'll also need to update calls to Parent#global_id to Parent#id, see: https://github.com/search?q=repo%3Asul-dlss%2Fvt-arclight%20global_id&type=code. We will also want to remove the line that sets the parent_ids attribute to parent_ssim and let the arclight default take over, see: https://github.com/sul-dlss/vt-arclight/pull/607/files#diff-6d990fa630d3862542f2ca619e89428c374b4ba000e78fb7783fee6386dfbd84R21
Cory Lown commented
Couple other things to figure out:
-
How long does a full re-index take? (Try on stage?)will handle as part of #627 - Can the app be set up to work with both the existing 1.0 index and an updated 1.1 index?
Cory Lown commented
Detailed todos for backward/forward compatibility:
- Change summary field config from
config.add_summary_field 'creators_ssim', label: 'Creator'
toconfig.add_summary_field 'creator_ssim', label: 'Creator'
- creators_ssm will no longer exist after re-index do we have to remove this from the requesthandler fl config or is this okay to leave until later.
-
date_range_ssim
changes todate_range_isim
. vt-arclight usesdate_hierarchy_ssim
. Only concern is do we need to remove date_range_ssim from the facet.field config. -
did_note_tesim
is removed in 1.1 do we need to remove the copy field directive totext
- Try 1.1 branch with the old index
- Try 1.1 branch with the new index
- Extra dot separator under title in results view after reindex; does removing extent index config break anything pre-reindex
- Need to try gsub approach to extent prior to reindex
- Empty date field showing up after reindex; somtimes normalized_date_ssm contains an empty string "id":"mt839rq8746aspace_d9f38b15d8b1cd785711604d2da730b4", "normalized_date_ssm":[""]
- Remove extent pill after title; we might want to revert to 1.0 extent indexing behavior to avoid weird stuff around this