Do not fail SUBSCRIBEs that start in the past #652
Conversation
Fixes #595 It's difficult or possibly even impossible to prevent it.
ianswett
added
the
Subscribe
ianswett
requested review from
kixelated,
suhasHere and
vasilvv
as code owners
|
Contrary to the title, this PR is not allowing a SUBSCRIBE to start in the past. It is taking a SUBSCRIBE that wants to start in the past, and coercing it to instead start at the beginning of the current group? |
ianswett
changed the title
Good point will, I retitled this PR. |
draft-ietf-moq-transport.md
Outdated
| StartGroup is prior to the current group, the subscription starts at the | ||
| beginning of the current group. |
There was a problem hiding this comment.
Individual Comment:
I'd say if the specified start point is prior to the latest seen object, the subscription starts at the next object .
There was a problem hiding this comment.
+1 with @afrind with minor tweaks .. "If the specified StartGroup is prior to the latest seen object, then the subscriptions starts at the latest group and object seen at the publisher ( as if the filter was Latest Object)"
There was a problem hiding this comment.
I mention the Latest Object filter now. I think it might be worth filing an issue for exactly what Object is delivered first for the Latest Object filter.
|
I made some minor updates, but I think a larger change is likely required to clarify what Objects one expects to receive. #592 |
Fixes #595
It's difficult or possibly even impossible to prevent it.