Last modified: 2014-07-10 06:40:52 UTC
From https://gerrit.wikimedia.org/r/#/c/115965/: egrep '"selser".* 5", "' ./parserTests-blacklist.js: add("selser", "Comment semantics: unclosed comment at end 5", "<!--This comment will run out to the end of the document-->"); add("selser", "HTML-pre: 1. embedded newlines 5", "<pre>foo</pre>\n\n<pre>\nfoo\n</pre>\n\n<pre>\nfoo\n</pre>\n\n<pre>\n\nfoo\n</pre>"); Subbu: Weird: debug_selser.sh script passes this test. Looked into this a bit too, but it does fail reliably in parserTests. The diff is a newline inside one of the pres. This does not happen in http://parsoid.wmflabs.org/_rtselser/mediawikiwiki/User:GWicke/Test/embedded_nls however. Some issue about non-selser wt2wt mode leaving dirty separator state maybe? add("selser", "Sanitizer: Closing of closed but not open table tags 5", "Table not started"); add("selser", "References: 5. ref tags in references should be processed while ignoring all other content 5", "A <ref name=\"a\" />\nB <ref name=\"b\">bar</ref>\n\n<references\n\n<references>\n<ref name=\"a\">foo</ref>\nThis should just get lost.\n</references>"); Subbu: This looks like a real bug that needs investigation and fixing. add("selser", "Tables: 1d. No escaping needed 5", "foo"); add("selser", "RT-ed inter-element separators should be valid separators 5", "\n{|\n|- [[foo]]\n|}"); add("selser", "2. Ensure fostered text content is wrapped in spans (traps regressions around fostered marker on the span getting lost) 5", "\n{|\n|- || ||\n| a\n|}"); Subbu: Extra leading newline at start of document -- seems harmless, but could be investigated. add("selser", "Table in fosterable position 5", "{{OpenTable}}\n<div>\n{|\n|}\n");
add("selser", "RT-ed inter-element separators should be valid separators 5", "\n{|\n|- [[foo]]\n|}"); add("selser", "2. Ensure fostered text content is wrapped in spans (traps regressions around fostered marker on the span getting lost) 5", "\n{|\n|- || ||\n| a\n|}"); add("selser", "Table in fosterable position 5", "{{OpenTable}}\n<div>\n{|\n|}\n"); These look like newlines leaking out of the foster box as discussed in https://gerrit.wikimedia.org/r/#/c/115856/.
> add("selser", "References: 5. ref tags in references should be processed > while ignoring all other content 5", "A <ref name=\"a\" />\nB <ref > name=\"b\">bar</ref>\n\n<references\n\n<references>\n<ref > name=\"a\">foo</ref>\nThis should just get lost.\n</references>"); > > Subbu: This looks like a real bug that needs investigation and fixing. Started looking at this, and this is looking extremely familiar -- this is a dsr issue and has something to do with how references tag with embedded content is processed. I remember investgating this 2-3 months back and getting distracted and dropping the ball on it. Am continuing with this now and see what the specific issue was that got in the way then. With the bug report, less likely to forget about it if I get pulled into something else :).
*** This bug has been marked as a duplicate of bug 67787 ***
