Provide a generic transaction-oriented bulk job worker
ClosedPublic
Actions

Authored by epriestley on Jan 10 2018, 9:06 PM.

Details

Reviewers

amckinley

Maniphest Tasks

T13025: Turn bulk editor into a generic piece of EditEngine infrastructure with Custom Field support

Commits

rP6ef45d824593: Provide a generic transaction-oriented bulk job worker

Summary

Depends on D18806. Ref T13025. See PHI173. Currently, Maniphest bulk edits are processed by a Maniphest-specific worker. I want to replace this with a generic worker which can apply transactional edits to any object.

This implements a generic worker, although it has no callers yet. Future changes give it callers, and later remove the Maniphest-specific worker.

Test Plan

See next changes.

Diff Detail

Repository

rP Phabricator

Branch

bulk3

Lint

Lint Passed

Unit

Tests Passed

Build Status

Buildable 19035
Build 25681: Run Core Tests
Build 25680: arc lint + arc unit

Event Timeline

epriestley created this revision.Jan 10 2018, 9:06 PM

epriestley edited the summary of this revision. (Show Details)Jan 10 2018, 9:07 PM

epriestley added a task: T13025: Turn bulk editor into a generic piece of EditEngine infrastructure with Custom Field support.

Harbormaster completed remote builds in B19028: Diff 45245.Jan 10 2018, 9:07 PM

epriestley requested review of this revision.Jan 10 2018, 9:07 PM

epriestley added a child revision: D18863: Define bulk edits in terms of EditEngine, not hard-coded ad-hoc definitions.Jan 10 2018, 9:12 PM

amckinley accepted this revision.Jan 11 2018, 12:26 AM

amckinley added inline comments.

src/applications/transactions/bulk/PhabricatorEditEngineBulkJobType.php
33–35	It feels a little heavy-handed to make new a new `BulkTask` for every PHID. On the other hand, I have a hard time imagining bulk jobs with like 1m+ PHIDs, or otherwise scaling to the point where this would start to get painful.

This revision is now accepted and ready to land.Jan 11 2018, 12:26 AM

amckinley mentioned this in D18863: Define bulk edits in terms of EditEngine, not hard-coded ad-hoc definitions.Jan 11 2018, 12:34 AM

I think each actual edit normally writes at least 5 rows when it actually applies (offhand: transaction, feed story, mail message, notification row, Herald transcript) so the overhead of the bulk job should always be fairly low compared to the overhead of actually performing an edit, even though it's high relative to, say, holding a string in memory.