Serialisation of client capabilities (take 2)#3909
Conversation
36c12a7 to
997b33a
Compare
997b33a to
997dcc2
Compare
There was a problem hiding this comment.
it's not a big deal, ignore this if you like. but in general (a) i don't like single-character identifiers, even locally; (b) i want us all to avoid edit wars.
There was a problem hiding this comment.
You don't like single-character identifiers, but two-character ones are ok? That's a bit of a strange stance. But sure, I'll revert it.
Co-authored-by: Matthias Fischmann <mf@zerobuzz.net>
ff26b74 to
1ce2ed8
Compare
|
@pcapriotti , in an attempt to get the PR through the CI, I took the liberty to rebase the PR on top of |
|
As far as I can tell, now we're getting actual test failures: https://concourse.ops.zinfra.io/builds/49403780 |
Thanks for rebasing. Yes, that's right. It should be fixed now. |
Fix an oddity in the serialisation of capabilities in the Client structure.
This PR replaces #3873 and simplifies the implementation. No new types are added, and the difference in serialisation is obtained using the Versioned wrapper, which we are already using for versioned changes in serialisation.
The change in the internal API between brig and galley is not backward compatible, so this can cause temporary failures during upgrades. The invariant that brig's clients are a superset of galley's clients will be maintained.
Note: #3904 is a previous attempt at this fix, which overlooked the fact that the
Clientstructure is used in user events, so it was reverted.https://wearezeta.atlassian.net/browse/WPB-6400
Checklist
changelog.d