beam-commits mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Davor Bonaci (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (BEAM-97) Input watermarks can never be null
Date Sat, 05 Mar 2016 22:35:40 GMT

     [ https://issues.apache.org/jira/browse/BEAM-97?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Davor Bonaci updated BEAM-97:
-----------------------------
    Assignee: Mark Shields  (was: Davor Bonaci)

> Input watermarks can never be null
> ----------------------------------
>
>                 Key: BEAM-97
>                 URL: https://issues.apache.org/jira/browse/BEAM-97
>             Project: Beam
>          Issue Type: Bug
>          Components: sdk-java-core
>            Reporter: Mark Shields
>            Assignee: Mark Shields
>
> A null input watermark (eg the result of TimerInternals.currentInputWatermarkTime())
is currently interpreted to mean 'computation is still starting and watermark is not yet known'.
However, the Google runner was also returning null for 'computation has just restarted and
previous watermark has not been established'. As a result code such as WatermarkHold would
create data holds using the first interpretation which is unsound for the second interpretation.
> Instead, input watermarks should never be null so the interpretation is never ambiguous.
The Google runner is ready to support this.



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message