"AG" == Aryeh Gregor Simetrical+wikilist@gmail.com writes:
AG> The only question is whether the rows are kept in the text table.
Yes, for wikis that have been around since 1.8, run this simple test $ mysql MyDatabase <<\EOF SELECT COUNT(*) FROM archive WHERE ar_namespace = 8 AND ar_user_text = 'MediaWiki default'; SELECT COUNT(*) FROM logging WHERE log_namespace = 8 AND log_comment = 'No longer required'; EOF COUNT(*) 1518 COUNT(*) 1510
(Note you may have to say e.g., FROM wiki_archive depending on how your tables are named.)
Now allow us to examine a small, but "healthy" wiki, http://transgender-taiwan.org/ , which was born after the 1.8/1.9 boundary in the Mediawiki stratigraphic record ( http://en.wikipedia.org/wiki/Cretaceous%E2%80%93Tertiary_extinction_event ).
$ mysqlshow --count transgender|egrep text|\ archive|Rows | Tables | Columns | Total Rows | | archive | 15 | 7 | | text | 3 | 62 |
This small wiki escaped having 1500 dinosaur corpse rows being added to each of those two tables.
Sure, the sanitary landfill that our database has now become will still function fine, but doesn't it feel bad to know that at least for a smaller wiki the majority of the rows may very well just be junk from a chance mishap at the 1.8-1.9 boundary? Yes, counting bytes instead of rows might help one sleep at night.
Anyway, how would a WikiSysop who feels this is bad correct it for his own wiki? I hope the method I used at the beginning of this thread is safe and correct.