Parsoid's Cite extension sometimes produces different ids for the same <ref> source
Closed, ResolvedPublic
Actions

Assigned To

Authored By

	ssastry
	Feb 11 2014, 5:14 AM

Description

This was originally noticed in repeated round-trip tests:

Not sure why cite ids are not identical anymore when double rt-diff is applied on a page .. This has been causing spurious rt diffs in the last 2 (at least) rounds of rt-testing.

Details

Reference: bz61165

Related Objects
Search...

Status	Assigned	Task
Invalid	None	T93751 RFC: Next steps for long-term revision storage -- space needs, storage hierarchies
Resolved	• GWicke	T93779 Only store a new render of Parsoid HTML / data-parsoid revision if the content actually changed after a template update
Open	None	T93715 [EPIC] Make Parsoid HTML output completely deterministic
Resolved	• marcoil	T63165 Parsoid's Cite extension sometimes produces different ids for the same <ref> source

Event Timeline

• bzimport raised the priority of this task from to Low.Nov 22 2014, 2:58 AM

• bzimport added a project: Parsoid-Tests.

• bzimport set Reference to bz61165.

ssastry created this task.Feb 11 2014, 5:14 AM

similar to this? 7b003c9b1ed1290bf71759ecd16be699c9fa05a7

Doesn't seem to be. roundtrip-test.js uses processTopLevelDoc both times.

Check: http://parsoid.wmflabs.org:8001/result/51c71eb04fd3111e04702c3729118fd3cfd611a9/hiwiki/%E0%A4%AE%E0%A4%BE%E0%A4%A4%E0%A5%83%E0%A4%B5%E0%A4%82%E0%A4%B6_%E0%A4%B8%E0%A4%AE%E0%A5%82%E0%A4%B9_%E0%A4%B5%E0%A4%BE%E0%A4%8F

So, the 2nd parse is showing ref id 40.

Now, check this: http://parsoid.wmflabs.org/_rt/hiwiki/%E0%A4%AE%E0%A4%BE%E0%A4%A4%E0%A5%83%E0%A4%B5%E0%A4%82%E0%A4%B6_%E0%A4%B8%E0%A4%AE%E0%A5%82%E0%A4%B9_%E0%A4%B5%E0%A4%BE%E0%A4%8F

There are only 5 refs. So, it shouldn't be getting an id of 40 at all. So, that is prettttty odd!

With recent cite fixes, haven't seen this much .. But leaving this open just in case it pops up its head again.

This seems to have been fixed by the fix for bug 67237

We are seeing a lot of these spurious diffs again in recent roundtrip test runs.

Arlolra added a project: Parsoid.Nov 25 2014, 7:14 PM

Arlolra set Security to None.

ssastry moved this task from Needs Triage to Testing on the Parsoid board.Dec 20 2014, 4:16 AM

• marcoil moved this task from Testing to Needs Triage on the Parsoid board.Feb 13 2015, 12:50 PM

• marcoil renamed this task from RT-testing: spurious semantic diffs triggered by different ref id attributes in Cite to Parsoid's Cite extension sometimes produces different ids for the same <ref> source.Mar 25 2015, 6:04 PM

• marcoil claimed this task.

• marcoil raised the priority of this task from Low to High.

• marcoil updated the task description. (Show Details)

• marcoil added a parent task: T93715: [EPIC] Make Parsoid HTML output completely deterministic.

• marcoil mentioned this in T93715: [EPIC] Make Parsoid HTML output completely deterministic.

Change 199643 had a related patch set uploaded (by Marcoil):
T63165: Ensure reference index is reset at the end of document

https://gerrit.wikimedia.org/r/199643

gerritbot added a project: Patch-For-Review.Mar 25 2015, 6:06 PM

Change 199643 merged by jenkins-bot:
T63165: Ensure reference index is reset at the end of document

https://gerrit.wikimedia.org/r/199643

ssastry mentioned this in T93973: Technical Debt: Eliminate all state from Cite.references object.Mar 25 2015, 11:31 PM

ssastry mentioned this in T93974: Tech Debt: Allocate native extension objects once per document instead of reusing it across documents.Mar 25 2015, 11:33 PM