Don't delete devices marked as dehydrated devices #4268
Conversation
uhoreg
force-pushed
the
ignore_dehydrated_devices
branch
from
550a615 to
20e2e24
Compare
richvdh
added
the
Z-Build-Workflow
sandhose
added
Z-Build-Workflow
|
Opens #4269 |
crates/matrix-synapse/src/lib.rs
Outdated
| let existing_devices: HashSet<String> = body | ||
| .devices | ||
| .into_iter() | ||
| .filter(|d| d.dehydrated.is_none()) |
There was a problem hiding this comment.
we could also alternatively consider hiding dehydrated devices from the admin API, but in principle I think it does make sense to show them there, so happy with this
There was a problem hiding this comment.
that said, shouldn't we have d.dehydrated != Some(true) here, or make dehydrated: bool with a default of false? (probably in favour of the latter since that's the simplest and the semantics match)
There was a problem hiding this comment.
though the struct is serialised again, so Option<bool> makes sense unless we split the usage — I'll change the filter condition, push, merge
crates/matrix-synapse/src/lib.rs
Outdated
| let existing_devices: HashSet<String> = body | ||
| .devices | ||
| .into_iter() | ||
| .filter(|d| d.dehydrated.is_none()) |
There was a problem hiding this comment.
that said, shouldn't we have d.dehydrated != Some(true) here, or make dehydrated: bool with a default of false? (probably in favour of the latter since that's the simplest and the semantics match)
reivilibre
removed
the
Z-Build-Workflow
sandhose
added
A-Homeserver-Integration
4 participants
MAS tries to delete all the devices that it doesn't know about. This PR makes it ignore the dehydrated device (as reported by Synapse), so that it won't get deleted.
Companion PR: element-hq/synapse#18252
Part of a fix to element-hq/element-meta#2784