Give a route for retrying invites for users which may not exist #2434
Conversation
Fixes element-hq/element-web#7922 This supports the current style of errors (M_NOT_FOUND) as well as the errors presented by MSC1797: matrix-org/matrix-spec-proposals#1797
|
CI is failing due to lint errors fixed on experimental. To ease the merge conflict, I'd recommend pretending they don't exist for now. |
There was a problem hiding this comment.
This looks along the right lines although the retrying doesn't feel quite right. I'm not sure it makes sense to automatically retry any failed ones with those error codes as it's unlikely to help (could it cause it to retry in a loop?). Also just describing them as 'retries' feels misleading because it's not telling me that it will behave differently when I hit the 'try again' button (ie. if I saw that, I would assume mashing 'try again' was pointless as it was not going to make the user appear out of nowhere).
I think I would probably fix this by making the dialog specific to only profile lookup failures, ie. only display it for errors where our profile lookup phase failed, make the copy refer to it as inviting these users anyway and make the option, "alwaysInviteUsersWithNoProfile" or something (which also should still work for a future 'show question marks' version of this).
|
Loops shouldn't be possible as the inviting logic has been broken out in a way where it won't get caught by the stuff that says the lookups failed. I'll update the dialog, etc to care more about showing profile errors (non-fatal). The "retry" wording is indeed a lie as it doesn't truly retry - it skips a bunch of checks and goes ahead with the invite. |
Also handle profile errors better
|
(tests are still failing due to other lint errors fixed on experimental) |
Fixes element-hq/element-web#7922
This supports the current style of errors (M_NOT_FOUND) as well as the errors presented by MSC1797: matrix-org/matrix-spec-proposals#1797
Note to the reviewer: This is intentionally done against
developinstead ofexperimentalso it lands in the upcoming release. It will need merging toexperimentalif approved. Also sorry for the large diff - I refactored the actual invite logic out of_inviteMoreso I could reuse it.