[10:18:17] 11 [11:07:07] <[1997kB]> hey, what's the reason for this message - There seems to be a problem with your login session; this action has been canceled as a precaution against session hijacking. Please resubmit the form. [11:31:22] [1997kB]: it's because a "CSRF token" gets baked into the login form every time you load it. it's to prevent bad people from mimicking the login page -- they won't have that token so can't actually log in as you [11:32:29] you get that message if the token doesn't match what the server expects, so it generates a new token and ask you to try again [12:11:50] <[1997kB]> douglas: A user emailed said that they receiving this message on multiple login attempts on meta wiki. What could be the issue with them. I don't think they are in mimicked page. [18:58:18] [[Tech]]; SilkTork; /* Problems with editnotice (Group notice / Page notice) */ new section; https://meta.wikimedia.org/w/index.php?diff=19425857&oldid=19421999&rcid=14205801 [19:57:20] hi, does anyone here know when can rev_len be 0? i'm looking at the revisions of 2-3 articles , and rev_len for all of them is 0 [19:57:33] this is wrt the revision table in mediawiki [19:59:15] I guess the articles do have content? [19:59:29] are those on a wikimedia project? [20:00:14] yes, on enwiki, i'm looking at the data after downloading a dump, [20:00:54] which pages are you looking at? [20:01:41] select p.page_title, p.page_id, r.rev_id, r.rev_user, r.rev_timestamp, r.rev_comment, r.rev_len, r.rev_deleted, r.rev_minor_edit from page p join revision r on p.page_id=r.rev_page where p.page_title IN ("'Almond'", "'Army'", "'Apollo_Program'", "'Android_(robot)'", "'Ada_(programming)'") order by r.rev_timestamp DESC limit 10 [20:01:53] these are normal pages, nothing special [20:04:12] you have too many quotes there [20:04:28] and for more efficiency, you should add AND p.page_namespace=0 [20:04:34] so that you can use the index [20:04:50] looking at wmf, rev_len is not empty: [20:04:59] +-----------------+---------+-----------+----------------+---------+-------------+----------------+ [20:05:02] | page_title | page_id | rev_id | rev_timestamp | rev_len | rev_deleted | rev_minor_edit | [20:05:05] +-----------------+---------+-----------+----------------+---------+-------------+----------------+ [20:05:08] | Army | 1376 | 918884471 | 20190930184518 | 28339 | 0 | 1 | [20:05:11] | Army | 1376 | 918883535 | 20190930184109 | 24996 | 0 | 0 | [20:05:14] | Army | 1376 | 918882958 | 20190930183842 | 27714 | 0 | 0 | [20:05:17] | Army | 1376 | 918882397 | 20190930183612 | 28277 | 0 | 0 | [20:05:20] | Army | 1376 | 918882245 | 20190930183529 | 28336 | 0 | 0 | [20:05:23] | Almond | 1064 | 918496540 | 20190928233723 | 58238 | 0 | 0 | [20:05:26] | Almond | 1064 | 918496348 | 20190928233609 | 58247 | 0 | 0 | [20:05:29] | Android_(robot) | 713 | 917854604 | 20190925201318 | 27123 | 0 | 1 | [20:05:32] | Almond | 1064 | 917804629 | 20190925153501 | 58238 | 0 | 1 | [20:05:35] | Army | 1376 | 917443698 | 20190923211534 | 28339 | 0 | 1 | [20:05:38] +-----------------+---------+-----------+----------------+---------+-------------+----------------+ [20:05:51] thanks for the full fax throwback experience [20:06:48] hehe, you wanna a black fax, too? ;) [20:07:00] Platonides: thats strange, i've downloaded the dump from - https://dumps.wikimedia.org/enwiki/20190901/ "pages-meta-history1" don't think downloading a part of the dump can be an issue [20:07:20] codezee: maybe it's a problem with your importer? [20:07:29] let me check the dump [20:08:47] * Platonides is not seeing that field in the dump [20:08:53] i'm using the operations-dumps-import-tools - https://phabricator.wikimedia.org/source/operations-dumps-import-tools/browse/master/xmlfileutils/ but i'll still take a look in the dump [20:08:55] should it be added by the importer? [20:12:07] Platonides: yes, looking like the xml dump does not contain bytes of the revision, so i see no way for the importer to pick that up [20:12:14] *looks like [20:13:28] well, the importer should do a strlen() [20:13:50] probably a bug? [20:14:17] looks like an importer bug, yes [20:14:38] I'm more worried about the quotes in the pagenames, though [20:15:14] yes, those are also there, was surprised to see them, but i thought it was a requirement [20:15:24] *part of a standard [20:15:47] no, they shouldn't be there [20:15:59] in fact, I had to remove them in tools for it to work [20:25:57] i also cannot directly get the textlen from the text table for each revision because the slots table creates an indirection and its not in the dumps i think :( [20:38:53] ~``2`2 [20:38:59] whoops, sorry [20:43:54] well, it depends [20:46:10] if it is a basic table, you can do: [20:46:19] select LENGTH(old_text) from revision join text on (rev_text_id=old_id) where rev_id= $N ; [20:46:41] however, it is also possible that the text table only has placeholders to external storage [20:47:04] or that multiple revisions were joined and compressed into a single blob [20:47:51] most likely, importDump create the basic format, and it will work