Importing an article from different language Wikipedia displays incorrect byte sizes for diffs in History
Closed, ResolvedPublic
Actions

Assigned To

Authored By

	doctaxon
	Oct 6 2015, 9:04 PM

Description

Hi! The problem is about the revision history of imported pages across wikimedia projects in all languages.

For example:

https://de.wikipedia.org/w/index.php?title=Silver_Falcons&action=history

as this entries:

(cur | prev) 2014-11-01T16:12:55‎ Farawayman (Talk | contribs | block)‎ . . (11,474 bytes) (+6,296)‎ . . (→‎Current team: added "outdated" banner to section) (undo | thank)
(cur | prev) 2014-03-27T15:01:28‎ Nathan121212 (Talk | contribs | block)‎ . . (11,449 bytes) (+6,271)‎ . . (→‎Team 73) (undo | thank)
(cur | prev) 2014-03-24T05:49:59‎ NJR ZA (Talk | contribs | block)‎ . . (11,466 bytes) (+6,288)‎ . . (removed Category:Aerobatic teams; added Category:South African aerobatic teams using HotCat) (undo | thank)
(cur | prev) 2014-03-20T20:05:10‎ Nathan121212 (Talk | contribs | block)‎ . . (11,452 bytes) (+6,274)‎ . . (undo | thank)
(cur | prev) 2014-03-17T16:03:27‎ Nathan121212 (Talk | contribs | block)‎ . . (11,416 bytes) (+6,238)‎ . . (undo | thank)
(cur | prev) 2014-03-17T15:59:58‎ Nathan121212 (Talk | contribs | block)‎ . . (11,387 bytes) (+6,209)‎ . . (→‎Current team) (undo | thank)

Look at the revision sizes and the diff sizes and tell me please, how 11,416 bytes + 6,274 can be 11,452 bytes.

Is it a bug in the mediawiki software due to importing pages from somewhere else into existing articles?

Delete and undelete the article after the import cleans up the revision sizes in the history, but deletes resp. destroys logs in the logbooks, this is no solution of the problem.

Thank you very much ...

Details

	Subject	Repo	Branch	Lines +/-
	Set correct parentid on import	mediawiki/core	master	+15 -0

Customize query in gerrit

Related Objects

Mentioned In: T121743: Quotes needed for timestamps in PostgreSQL
T115975: Broken parent revisions on some old imported edits
T38976: Diffs: Incorrect number of bytes added or removed because rev_parent_id is set to wrong revision

Event Timeline

doctaxon created this task.Oct 6 2015, 9:04 PM

doctaxon assigned this task to Aklapper.

doctaxon raised the priority of this task from to High.

doctaxon updated the task description. (Show Details)

doctaxon added a project: MediaWiki-Core-Snapshots.

doctaxon added subscribers: doctaxon, Umherirrender, Luke081515 and 4 others.

Ireas subscribed.Oct 6 2015, 9:07 PM

Thgoiter subscribed.Oct 7 2015, 6:38 AM

Thanks for taking the time to report this!

Look at the revision sizes and the diff sizes and tell me please, how 11,416 bytes + 6,274 can be 11,452 bytes.

Looks like the calculation mistakenly refers to the revision from Oct 4th which has 5.178 bytes.
https://de.wikipedia.org/w/index.php?title=Silver_Falcons&type=revision&diff=146645073&oldid=146669452 is the import from en.wp presumably.

(Resetting priority of this task to its default - I don't see yet how this it's an urgent problem and why it is important for your work to have correct diff byte sizes for the important English versions but feel free to elaborate)