This is a followup of #1194136: Re-organise core cache bins, comment #106.
Problem/Motivation
The only problem left after #1194136: Re-organise core cache bins is that core provides also a non-bin cache table: cache_tags
. And that has the same name pattern as the bins. If you manually truncate the DB cache you may accidentally truncate also cache_tags
(I did it!).
Proposed resolution
DB cache bins should have a different name pattern than the cache tags table. Keep them as {cache_*}
and rename {cache_tags}
to {cachetags}
. Cache tables naming will be like:
- cache_bootstrap
- cache_data
- ...
- cache_render
- cache_toolbar
- cachetags
And so bins will have their own namespace.
Remaining tasks
Agreement, patch, RTBC, commit.
User interface changes
None.
API changes
None.
Comment | File | Size | Author |
---|---|---|---|
#8 | 2302843-8.patch | 6.64 KB | claudiu.cristea |
Comments
Comment #1
claudiu.cristeaPatch.
Comment #2
claudiu.cristeaComment #3
claudiu.cristeaComment #4
claudiu.cristeaComment #5
sunAdjusting issue title, because I first misunderstood what this is about.
This related issue still intends to refactor cache tags into a separate service, which may or may not change its db table.
Instead of changing all cache bin table names, we could also consider to change the table name of cache tags - to match their service names:
Comment #6
dawehnerThe proposal by sun would make it 100% clear that the cachetags table is a special snowflake, not a regular cache table.
The length of the tablenames would also be shorter which is nicer to look at.
Comment #7
claudiu.cristeaWell, I agree the design proposed by @sun. In fact both are doing the job but I went with my approach to have a minimal patch. Now
{cache_tags}
has more occurrences. Let me get back with a patch shortly.Comment #8
claudiu.cristeaWhile we don't want to provide an update path from
{cache_tags}
to{cachetags}
I think this should be committed before BETA. I don't think I'm allowed to tag, so I'm asking the reviewer to do it.Patch!
Comment #9
claudiu.cristeaUpdated also the issue summary.
Comment #10
claudiu.cristeaComment #11
dawehnerCool
Comment #12
Wim Leers+1 for doing the simplest thing possible.
Comment #13
alexpottCommitted db72be7 and pushed to 8.x. Thanks!