Pending queries not notified on timeout (Apollo)

[arjan]

It seems that when a query takes longer than 10 seconds, the phoenix channel times out (10s is the default timeout when expecting a response to .push()).

When this happens, the query keeps being stuck in the networkStatus = 1 state (request pending), instead of that an error is pushed up the Apollo link chain.

I tracked it down to the pushRequestUsing function, where the onTimeout handler calls notifyActive(), whereas the onError handler calls abortNotifier. When I change the onTimeout to call abortNotifier(), all works as expected.

Is this something worth PR'ing or is there a different approach?

[slorber]

Hi,

Do you think this is the same errur as this stacktrace?

AppApolloClient.tsx:230 createAbsintheSocketLink onError Error: channel join: timeout
    at createChannelJoinError (joinChannel.js:10)
    at joinChannel.js:16
    at Array.forEach (<anonymous>)
    at notifyErrorToAllActive (joinChannel.js:13)
    at Object.onTimeout [as callback] (joinChannel.js:31)
    at phoenix.js:1
    at Array.forEach (<anonymous>)
    at e.value (phoenix.js:1)
    at Object.callback (phoenix.js:1)
    at e.value (phoenix.js:1)