This patch implements caching for COBs by using an SQLite database that is used as a write-thru cache. All reads are replaced with using the cache so as to avoid recomputing the COB graph each time.
This patch implements caching for COBs by using an SQLite database that is used as a write-thru cache. All reads are replaced with using the cache so as to avoid recomputing the COB graph each time.
Changes:
- added
cachesubcommands forpatchandissue
Changes:
- Document modules and new additions
- Order by is listing queries
- Add listing by different statuses for patches (includes adding new Status type)
- recache split into two functions, write and write_all
- reader db’s no longer migrate
- formatted sql schemas
Remove::removeno longer required the RID- Clean up of PatchMut
- cache/cobs.db renamed to cobs/cache.db
- Remove Cacheable type and inline usage of its methods
- Simplify Cache types to take any R, not necessarily the underlying
stores. This allows for more flexible caching and definitions,
e.g. when paired with anything that implements
HasRid(a newly introduced trait), we get all of thePatchesandIssuesmethods. This could potentially mean we could unify the two differentCachetypes in a future change.
N.B. all new commits are fixups to help with revision-to-revision review and will be squashed once it’s ready for merging.
Changes:
- Removed Filter type in patch command
- Fixed COB type name in update error
- Squashed fixup commits
Rebase.
Fix broken build due to bad import.
Reset to master.
@fintohaps I did not look into the actual changes yet, before I do that I would like get clarification from you that this patch is indeed still relevant. It’s been a while… Thanks!