Page MenuHomePhabricator

Pholio - fix show older transactions for if there are inline comments
ClosedPublic

Authored by btrahan on Dec 17 2014, 5:07 PM.
Tags
None
Referenced Files
F14053850: D11002.diff
Fri, Nov 15, 8:02 PM
F14041608: D11002.diff
Mon, Nov 11, 7:59 PM
F14025572: D11002.diff
Thu, Nov 7, 6:52 PM
F13999581: D11002.diff
Thu, Oct 24, 3:45 PM
F13993889: D11002.id26417.diff
Wed, Oct 23, 2:03 AM
F13993740: D11002.diff
Wed, Oct 23, 1:06 AM
F13978754: D11002.id26416.diff
Oct 19 2024, 12:48 AM
Unknown Object (File)
Sep 17 2024, 7:36 AM
Subscribers

Details

Summary

We didn't load enough data for this case for the custom view class Pholio uses. Fixes T6770. Re-jiggers the signature of the loadImages function in the PholioMockQuery to get there so as to not duplicate any business logic.

Test Plan

made a pholio mock with lots of inline comments. pre-patch "show older" fatals and post-patch "show older" works

Diff Detail

Repository
rP Phabricator
Lint
Lint Not Applicable
Unit
Tests Not Applicable

Event Timeline

btrahan retitled this revision from to Pholio - fix show older transactions for if there are inline comments.
btrahan updated this object.
btrahan edited the test plan for this revision. (Show Details)
btrahan added reviewers: epriestley, chad.
epriestley edited edge metadata.

I wonder if we should eventually formalize the idea of "attachment points" and provide some more standard way to fill in missing attachments. The attachX() + getX() + needX() approach generally seems to work pretty well, but it definitely falls down a bit in cases like this.

We had some ad-hoc loadAndAttach() type things before, but they were all custom and nonstandard and operated on only a single object. I also don't like the`loadRelatives()` approach much -- it always felt too magical and lacked power once we implemented policies.

If "attachable" properties were slightly more formal, Query classes could potentially define how to load each type of attachment and then a common API could say "load attachment X for these objects" and be used on both the primary Query pathway and cases like this where we need to fill data in later.

Anyway, not a big deal (this doesn't come up often) and I don't have any concrete ideas on an implementation yet, but it's something I imagine maybe coming up with an approach for in the next year or so.

This revision is now accepted and ready to land.Dec 17 2014, 5:17 PM
This revision was automatically updated to reflect the committed changes.