Open vaudevillen opened 4 years ago
From a philosophical standpoint, a system should be considered "in error" for the entire time that it cannot acquire a new value. Therefore the only "true" way to clear an error is to execute a successful fetch.
Right now I'm leaning toward restricting any external control over errors because it might confuse or dilute their meaning.
Because
eachError
is backed by aConflatedBroadcastChannel
, it'll replay errors, but I might not be concerned with any previous error values. It would be nice if we could either clear the error, or have this work like aflow
and "drop" the first error. Also, this seems related to issue https://github.com/asarazan/bismarck/issues/13, and could be solved by adding the ability to manually update the error value