You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When aligning / joining multiple streams it's important to know when the streams are up to date with each other. Currently this is not possible [1^].
It would be ideal if the up-to-date message had some form of LSN header that provided some sort of assurance of there this stream is up to date to.
There is likely to be some complexity in providing this due to the concurrent processing on shapes by a single server. Essentially it needs to be a message that says that this shape will not receive any messages <= a specific LSN, but can't just be calculated from the last LSN seen for a particular shape. It needs to be aware of the lower bound of LSNs that have been received and filtered.
[1]: We can sort of infer it is by aborting a long poll, reconnecting with live=false and as the request was made after the last message on another stream we can "guess" that we are up to date if we get an up-to-data message. However there is no guarantee (shapes are processed concurently), and we don't know for sure at what point that up-to-date was sent, so if any message came in in the interim it's impossible to know where it sits.
The text was updated successfully, but these errors were encountered:
When aligning / joining multiple streams it's important to know when the streams are up to date with each other. Currently this is not possible [1^].
It would be ideal if the
up-to-date
message had some form of LSN header that provided some sort of assurance of there this stream is up to date to.There is likely to be some complexity in providing this due to the concurrent processing on shapes by a single server. Essentially it needs to be a message that says that this shape will not receive any messages <= a specific LSN, but can't just be calculated from the last LSN seen for a particular shape. It needs to be aware of the lower bound of LSNs that have been received and filtered.
[1]: We can sort of infer it is by aborting a long poll, reconnecting with
live=false
and as the request was made after the last message on another stream we can "guess" that we are up to date if we get anup-to-data
message. However there is no guarantee (shapes are processed concurently), and we don't know for sure at what point that up-to-date was sent, so if any message came in in the interim it's impossible to know where it sits.The text was updated successfully, but these errors were encountered: