Community Wishlist Survey 2019/Miscellaneous/Recover text of old revisions for the Massachusetts article

Random proposal ►◄ Miscellaneous  The survey has concluded. Here are the results!

Recover text of old revisions for the Massachusetts article

• Problem: The deletion and undeletion events from 2005 were terrible events that have caused lots of old revisions from January 2005 and earlier for the Massachusetts article to have lost their text. This is a catastrophe that had never been fixed for over 13 years.

• Who would benefit: Everyone

• Proposed solution: Download some database dump from February 2005 and extract the Massachusetts article edits. Then create a temporary page (perhaps a talk subpage) containing the fact that the text had been lost, and use Special:MergeHistory to merge all of the article edits from 7 January 2005 and earlier to the temporary page's history. Next, delete the temporary page and restore the latest creation as well as the few earlier non-problematic revisions. After that, merge the earlier non-problematic revisions back to the Massachusetts article's history. Then, import all the old Massachusetts article edits from the database dump. And optionally, create a maintenance script that attempts to copy the text of the imported revisions to the corresponding deleted revisions for the temporary page, and rev_len and rev_sha1 to ar_len and ar_sha1 respectively. Unfortunately, this is more complicated due to the fact that MCR means that we are not going to have rev_text_id and ar_text_id fields anymore. Finally, re-delete the temporary page (optionally also).

• More comments:

• Phabricator tickets: T147146

• Proposer: GeoffreyT2000 (talk) 16:29, 30 October 2018 (UTC)

Discussion

• This server crash? What data did you lose? HLHJ (talk) 02:51, 18 November 2018 (UTC)
  • 95% editions before mentioned date are inaccessible. Ankry (talk) 04:16, 18 November 2018 (UTC)
• This is assuming that backups from between late 2003 and before Jan 2005 exist right ?.. Because I'm not so sure they do :) Maybe @Tim Starling (WMF): has some tucked away. —TheDJ (talk • contribs) 14:20, 19 November 2018 (UTC)
  • I didn't have the disk space back then to tuck away gigabytes of dumps, but I see Internet Archive has an SQL dump from December 2004: https://archive.org/download/20041223_old_table.sql -- Tim Starling (WMF) (talk) 04:42, 20 November 2018 (UTC)

Voting

•  Support Liuxinyu970226 (talk) 08:43, 17 November 2018 (UTC)
•  Support Ankry (talk) 04:15, 18 November 2018 (UTC)
•  Support Sebastian Wallroth (talk) 13:11, 18 November 2018 (UTC)
•  Support Why hasn't this been done before? Benjamin (talk) 10:32, 19 November 2018 (UTC)
•  Support Novak Watchmen (talk) 01:34, 21 November 2018 (UTC)
•  Support Terra ❤ ^(talk) 02:12, 21 November 2018 (UTC)
•  Support Vulphere 04:44, 21 November 2018 (UTC)
•  Support Lotje (talk) 06:47, 25 November 2018 (UTC)
•  Support PMG (talk) 17:07, 26 November 2018 (UTC)