[10:50:10] <taavi>	 kostajh: thanks for the review! can you also look at https://gerrit.wikimedia.org/r/c/mediawiki/extensions/OATHAuth/+/988413/ for the same bug?
[16:10:18] <p4r4D0xum>	 hello, I try to setup mediawiki-1.40 on apache-2.4 
[16:10:57] <p4r4D0xum>	 Can't successfully run a setup, due to Runtime error.
[16:11:05] <Reedy>	 Which is? :)
[16:11:15] <p4r4D0xum>	 Reedy: Getting to that :)
[16:11:54] <p4r4D0xum>	 Mediawiki fails to find writable tmp directory
[16:12:08] <p4r4D0xum>	 No writable temporary directory could be found...
[16:12:35] <p4r4D0xum>	 Did I miss something in httpd.conf?
[16:12:48] <p4r4D0xum>	 orrrrrr
[16:13:13] <p4r4D0xum>	 the default is '/tmp' and it can't be accessed by 'http' user?
[16:13:28] <p4r4D0xum>	 butttt
[16:13:30] <Reedy>	 More likely something in your PHP config (and/or the defaults don't work quite right)
[16:14:47] <Reedy>	 !phpinfo
[16:14:48] <wm-bot>	 create a file called info.php in your wiki's installation directory. It should have the following content: <?php phpinfo(); ?>; Visit it with your browser to see information about your php installation. If it doesn't work, check file permissions. Remove the file when done, it makes life easier for attackers.
[16:20:20] <p4r4D0xum>	 Reedy: I've got it.
[16:21:05] <p4r4D0xum>	 Reedy: sys_tem_dir? no value
[16:21:22] <p4r4D0xum>	 s/_tem_/_temp_
[16:53:35] <p4r4D0xum>	 fffff yeah!
[16:53:55] <p4r4D0xum>	 It was actually php_admin issue
[17:24:13] <cariaso>	 https://patientuser.com/deep/wiki/Annotation:HLA-B shows the text I expect, but SELECT p.page_title, t.old_text FROM page as p, text as t where t.old_id == p.page_latest and p.page_title = "Annotation:HLA-B"; shows different text. can anyone help me to understand why?
[17:31:50] <Reedy>	 bypassing the revision table?
[17:33:07] <cariaso>	 that query seemed to work well until a recent DB hiccup. Can you point me at a more correct sql query to retrieve latest page text?
[17:33:50] <Reedy>	 Probably more by luck than judgement
[17:34:56] <Reedy>	 > To retrieve the text of an article, MediaWiki first searches for page_title in the page table. Then, page_latest is used to search the revision table for rev_id, and rev_text_id is obtained in the process. The value obtained for rev_text_id is used to search for old_id in the text table to retrieve the text. When a page is deleted, the revisions are moved to the archive table.
[17:43:02] <moonmoon>	 (plus additional complications on wikis where the text table is hosted externally or page content is stored elsewhere via other means)
[17:49:55] <cariaso>	 I'm still poking at it, but I expected this to be right:
[17:49:56] <cariaso>	 select p.page_title, t.old_text from page as p, revision as r, text as t where p.page_title = 'Annotation:HLA-B' and p.page_id = r.rev_page and t.old_id = r.rev_text_id
[17:49:56] <cariaso>	 however it seems rev_text_id is deprecated, and now there is some path via either slots or content? can anyone get me closer?
[17:52:54] <Vulpix>	 from slots, slot_revision_id is the revision id. slot_content_id should join with content (content_id). And content_address will give you the id for the text table
[18:27:30] <cariaso>	 Vulpix I've got a larger query that's failing to return anything, and this seems to isolate why.
[18:27:31] <cariaso>	 select * from content, text where content.content_address = text.old_id;
[18:27:31] <cariaso>	 returns nothing. Any thoughts on why?
[18:28:37] <cariaso>	 ah, I see there is a "tt:" prefix
[18:29:01] <cariaso>	 which seems to preclude a simple join
[18:29:32] <cariaso>	 so I think this would be right except for that
[18:29:34] <cariaso>	 select page.page_title, text.old_text
[18:29:34] <cariaso>	 from page, revision, slots, content, text
[18:29:35] <cariaso>	 where page.page_id = revision.rev_page
[18:29:35] <cariaso>	 and revision.rev_id = slots.slot_revision_id
[18:29:36] <cariaso>	 and slots.slot_content_id = content.content_id
[18:29:36] <cariaso>	 and content.content_address = text.old_id
[18:30:22] <cariaso>	 I find it hard to believe I ought to be doing string manipulations for such a core functionality. Am I missing something in how this content table should be used?
[18:31:34] <Reedy>	 What are you actually trying to do?
[18:31:58] <Reedy>	 There's plenty of wrapper classes (or APIs) you shouldn't (probably) be needing to do SQL queries yourself
[18:31:59] <cariaso>	 I want to ready wikitext content for a given page from sql
[18:32:19] <cariaso>	 I need to walk all pages
[18:32:57] <cariaso>	 so apis would trigger N queries and would be drastically less efficient
[18:33:20] <Reedy>	 Many API calls can give you multiple results
[18:34:13] <cariaso>	 I'm on the server. I have direct access to the sqlite file. the apis are just overhead.
[18:34:50] <Reedy>	 so is potentially maintaining a query through schema changes :)
[18:34:57] <cariaso>	 I accept that
[18:35:46] <cariaso>	 I have very few queries I care about, it's really just  title --> wikitext
[18:39:45] <Reedy>	 SUBSTRING(content_address, 1, 3)
[18:42:11] <cariaso>	 select page.page_title, text.old_text
[18:42:12] <cariaso>	 from page, revision, slots, content, text
[18:42:12] <cariaso>	 where page.page_id = revision.rev_page
[18:42:13] <cariaso>	 and revision.rev_id = slots.slot_revision_id
[18:42:13] <cariaso>	 and slots.slot_content_id = content.content_id
[18:42:14] <cariaso>	 and SUBSTRING(content_address, 4) = text.old_id
[18:42:15] <cariaso>	 does it
[18:42:26] <cariaso>	 many thanks to Reedy and Vulpix
[18:46:04] <cariaso>	 and moonmoon
[19:36:50] <Platonides>	 that will break in case of external storage
[19:37:03] <Platonides>	 but I don't think you're going to use it on that server
[19:37:08] <Reedy>	 I imagine if they're using sqlite...
[19:37:09] <Reedy>	 :)
[19:37:27] <Reedy>	 of course, with ES they couldn't do cross server joins anyway
[19:58:29] <Platonides>	 yep, that was the reason I thought that ;)
[20:05:01] <Vulpix>	 You also won't get text if you compress revisions, which may be more common than setting ES