9c8d5e40ad
Always apply patches on load
2025-11-21 09:15:19 +00:00
f665b5a1fa
No need for default schema
2025-11-21 09:15:19 +00:00
4f6ec19491
Drop comments
2025-11-21 09:15:19 +00:00
0e11a139a5
Move quoting
2025-11-21 09:15:19 +00:00
62d102578f
Table name already correct
2025-11-21 09:15:19 +00:00
ca17193620
Belongs_to not optional
2025-11-21 09:15:19 +00:00
6dfb961b83
Delegate to_s to terms
2025-11-21 09:15:19 +00:00
ec06b7085b
No need to default created_at
2025-11-21 09:15:19 +00:00
5edb0a2075
Update schema
2025-11-21 09:15:19 +00:00
19b0e25eac
Split sqlite/mysql record stuff
2025-11-21 09:15:19 +00:00
a4850fbe73
Remove adapter specific logic from searchable
2025-11-21 09:15:19 +00:00
a682a807ed
Handle SQLite date subtraction
2025-11-21 09:15:19 +00:00
cc7e091508
Disable multi-db for SQLite
2025-11-21 09:15:19 +00:00
bb36b4846f
Support SQLite searching
2025-11-21 09:15:19 +00:00
a2333d9a37
Add SQLite support
...
- UUID support
- Schema compatibility layer, ignore MySQL specific options
- Search not working yet
2025-11-21 09:15:19 +00:00
7f76835150
Merge pull request #1668 from basecamp/pinned-stale
...
Several fixes for pinned cards
2025-11-21 10:03:32 +01:00
9d08f26c59
Disable turbo when navigating to another account
...
So that the turbo-permanent trays get properly replaced
https://app.fizzy.do/5986089/cards/3094
2025-11-21 09:54:16 +01:00
021831f810
Extract partials for the different sections
2025-11-21 09:48:00 +01:00
85b9e3dfd4
Make the notifications tray reload on broadcasted refreshes
...
We can't use the normal refresh: :morph because everything is contained in a turbo permanent section
so that the trays persist across regular navigation.
2025-11-21 09:41:03 +01:00
ae640198ae
Stick with the same general retention policy we use for everything
2025-11-21 09:40:50 +01:00
0a88565bd7
Slim down
2025-11-21 09:39:59 +01:00
f66da318aa
No longer needed now that we are back to a single DB
2025-11-21 09:39:33 +01:00
62d4998f27
We stopped using load_async in the code base
2025-11-21 09:35:25 +01:00
e813b7eaf0
Invalidate HTTP caching when the cards change
...
https://app.fizzy.do/5986089/cards/3067
2025-11-21 09:32:02 +01:00
1211e180c2
Merge pull request #1667 from basecamp/prevent-lexxy-dup
...
Fix: duplicated lexxy when toggling done while in edit mode
2025-11-21 09:03:09 +01:00
649888dbd4
Making the frame permanent with an id would reconnect the editor causing the dup
...
This is something we need to fix in Lexxy for good, but for now this patch will do it.
https://github.com/basecamp/lexxy/issues/263
2025-11-21 08:52:10 +01:00
c9e2927179
Merge pull request #1666 from basecamp/identify-account-in-notificaton-bundle
...
Show account if you have multiple
2025-11-20 22:05:33 -06:00
9bae30f771
Show account if you have multiple
...
It's hard to know which account email notifications are coming from
2025-11-20 22:03:54 -06:00
93375f59cd
Merge pull request #1665 from basecamp/magic-link-in-subject
...
Display code right in the subject line
2025-11-20 21:52:12 -06:00
dca002bd64
Update test
2025-11-20 21:51:07 -06:00
b1762ea6b6
Display code right in the subject line
2025-11-20 21:45:26 -06:00
5e18d79999
Merge pull request #1661 from basecamp/revert-1660-revert-1659-notifications-copy
...
Add more detail to events summaries
2025-11-20 21:41:09 -06:00
ad7eb594d3
Match copy coming in #1663
2025-11-20 21:31:40 -06:00
9ab15688c2
Merge pull request #1664 from basecamp/flavorjones/fix-activejob-retry
...
Make sure jobs are enqueued only after all transactions are committed
2025-11-20 19:07:20 -05:00
2c2a1c4b4a
Make sure jobs are enqueued only after all transactions are committed
...
This should fix the recurring ActiveStorage::FileNotFoundError errors
from ActiveStorage::AnalyzeJob.
The issue is that, when images are not direct uploaded (e.g. a card
background image or avatar image), the AnalyzeJob has been getting
enqueued before the file itself was uploaded to blob storage.
Setting ActiveJob::Base.enqueue_after_transaction_commit = true makes
sure that jobs are always enqueued only once *all* transactions have
been committed.
This will be the Rails default at some point, but for now we still
need to explicitly set it.
Big thanks to @jeremy for the assist. 👏
2025-11-20 18:47:42 -05:00
fdee7aaaa5
Merge pull request #1654 from basecamp/flavorjones/3063-activity-spike-deadlocks
...
Fix race condition in Card::ActivitySpike::DetectionJob
2025-11-20 15:53:47 -05:00
1bbb6d0c03
Fix race condition in Card::ActivitySpike::DetectionJob
...
The check-then-act pattern in `register_activity_spike` has been
replaced with `find_or_create_by!` to eliminate the race condition
that could lead to creating multiple activity spikes for a card. To
support this change, the `card_id` index on `cards_activity_spikes`
has been made unique.
ref: https://app.fizzy.do/5986089/cards/3063
2025-11-20 15:40:42 -05:00
c052839bd6
Removed quotes since we were not using them
2025-11-20 21:37:41 +01:00
a5b58e47fe
Remove unclosed span
2025-11-20 21:32:43 +01:00
baadd484b4
Review triage
2025-11-20 21:28:41 +01:00
20020043f1
Merge pull request #1662 from basecamp/flavorjones/fix-up-fk-migration
...
Make sure the migration to remove FK constraints is robust
2025-11-20 15:20:50 -05:00
fb742d6d29
Make sure the migration to remove FK constraints is robust
...
where we've already removed the constraints.
2025-11-20 15:16:26 -05:00
3f21ea2329
Revert "Revert "Add more detail to events summaries""
2025-11-20 21:12:41 +01:00
fd812cf862
Merge pull request #1660 from basecamp/revert-1659-notifications-copy
...
Revert "Add more detail to events summaries"
2025-11-20 21:12:37 +01:00
7200af879f
Revert "Add more detail to events summaries"
2025-11-20 21:12:10 +01:00
4aef7230e5
Merge pull request #1659 from basecamp/notifications-copy
...
Add more detail to events summaries
2025-11-20 21:11:58 +01:00
a48aa84f11
Remove legacy event
2025-11-20 21:11:02 +01:00
93a80ab8c2
Add more detail to events
2025-11-20 21:07:55 +01:00
4804e1367c
Merge pull request #1658 from basecamp/fizzy-remove-foreign-key-constraints
...
Drop all foreign key constraints
2025-11-20 15:06:00 -05:00
6e67a7787a
Drop all foreign key constraints
...
These have been a contributing cause to some deadlocks, but also this
makes some operational tasks harder (data loading, data truncation)
and has some performance impact on insert/update/delete operations.
The team consensus is that we aren't relying on them, and so we're
comfortable with the referential integrity risk.
2025-11-20 15:01:25 -05:00
Donal McBreen