GitHub user Xiao-zhen-Liu added a comment to the discussion: Proposal: operators that emit on more than one output port
Thanks for splitting the result-handling changes into small PRs, @sshiv012. Following up on our meeting on Friday, let us take this in two phases: 1. **Finish multi-port result support for native operators (`Split`, `If`) first:** - I will review #9271 and #9272 first so that `WorkflowResultStore`, `operator_port_executions` size and count tracking, and `WebResultUpdateEvent` handle all external output ports rather than only port 0. (Please rebase #9271 onto current `main`, since #9049 recently updated `onResultUpdate` in `ExecutionResultService.scala` and caused a merge conflict.) - For PR 3 (UI tabs per port), please coordinate with @kz930's #9052 series: #9063 already adds per-port tabs in the pinned result windows on the canvas, and #9060 removes the Result tab from the bottom panel. Once #9272 lands, the pinned window tabs in #9063 can use the per-port updates from `WorkflowResultService` directly instead of re-polling port 1 on a timer. - PR 4 (adding `portIndex` to `ResultExportRequest` and `ResultExportService`) is also needed for native multi-port operators, since `ResultExportService.getOperatorDocument` still hardcodes port 0 (which also forced #9063 to hide cell download on ports past the first). 2. **Discuss and finalize the Python UDF multi-port design before opening PRs 5 and 6:** - Supporting multiple output ports on Python UDFs touches port schema propagation, the `pytexera` user API, `pyamber` tuple routing, and Peanut's UDF compilation model. Let us have a dedicated design discussion to settle those questions and agree on the design before opening PRs 5 and 6. GitHub link: https://github.com/apache/texera/discussions/8914#discussioncomment-19500989 ---- This is an automatically sent email for [email protected]. To unsubscribe, please send an email to: [email protected]
