MZMcBride added a comment. > Today I discovered that our `ips_site_page` field is to small to store all > page titles out there (eg. https://www.wikidata.org/wiki/Q6703647 is affected > by that). The field is defined as `varbinary(255) NOT NULL` which would be > enough for the unqualified page titles (page_title is varbinary(255) > everywhere), but it's not large enough to hold the page titles plus their > namespaces (that's how we store the titles, unlike the `page` table which has > a separate int field for the namespace).
Is it a good idea for the `ips_site_page` field to store the namespace name as a string? Why is it doing that? > To make that field fit for all pages on every Wikimedia wiki we should go for > a length of at very least 300, as the longest page title I could find has 255 > chars and the longest non-talk namespace has a length of 39 (The longest talk > namespace has a length of 67!), also we need to store the colon. I'd suggest > to go for a bit more than that, just to play it safe. `page.page_title` is 255 bytes today, but I'm not sure it's a great idea to make other fields dependent on its length. Matching its length might be more reasonable. While I understand smaller fields are better for performance, I'd like if `page.page_title` and similar fields followed the model now being used with the *_comment fields (cf. https://phabricator.wikimedia.org/T6715). That is, have a higher limit on the database side and enforce the limit at the application level. From my reading of the numbers here, 300 bytes wouldn't be sufficient in the current most pathological case (255 bytes + 67 bytes). TASK DETAIL https://phabricator.wikimedia.org/T99459 EMAIL PREFERENCES https://phabricator.wikimedia.org/settings/panel/emailpreferences/ To: MZMcBride Cc: MZMcBride, Krenair, jcrespo, Springle, Lydia_Pintscher, aude, daniel, Aklapper, hoo, Wikidata-bugs _______________________________________________ Wikidata-bugs mailing list Wikidata-bugs@lists.wikimedia.org https://lists.wikimedia.org/mailman/listinfo/wikidata-bugs
