Last modified: 2014-01-12 20:18:12 UTC
1) Page sections containing UNIQ key can no longer be referenced directly by permanent/external links. 2) "API prop=sections" fails in section-titles containing UNIQ key, for example <s toclevel="1" level="2" line="UNIQ28cf615b3f10792f-nowiki-0000000B-QINU" number="39" /> where section-title-text behind the UNIQ key is lost: http://de.wikipedia.org/w/api.php?action=parse&page=Wikipedia:Fragen_zur_Wikipedia/Archiv/2009/Woche_10&prop=sections
UNIQ strings are not a feature, they are a consequence of malformed parsing (basically a big flashing sign that says "BAD WIKICODE HERE". In this case it seems no valid anchor string can be formed from [[Link|<nowiki>text</nowiki>]]. Changing topic appropriately. This might be a dupe or WONTFIX though.
(In reply to comment #1) OK, thanks. What puzzles me is, why do UNIQ strings have to change at all and/or so frequently? And even when a wiki-page doesn't change? That's a real drag in archived talk-pages, where "nowiki" is used quite frequently in section-titles.
Created attachment 5983 [details] Proposed fix: Swap the order of tag and link unstripping in Parser::formatHeadings() Easy fix: Swap the order of link unstripping and extension tag unstripping in Parser::formatHeadings(). It is IMHO far more likely that a link contains an extension tag (like above), while offhand I can't think of a case, where an extension tag would contain an half-parsed link at that stage of parsing. (In reply to comment #2) > (In reply to comment #1) OK, thanks. What puzzles me is, why do UNIQ strings > have to change at all and/or so frequently? And even when a wiki-page doesn't > change? That's a real drag in archived talk-pages, where "nowiki" is used quite > frequently in section-titles. > The whole point of UNIQ keys is that they are not predictable, so you can't break parsing by inserting some of them to a page.
related: bug 25417 (<ref> used also strip markers inside wikilink)
(In reply to comment #3) > Created attachment 5983 [details] > Proposed fix: Swap the order of tag and link unstripping in > Parser::formatHeadings() Paul, could you apply the patch now that you have commit privs?
Looks like this is already fixed from my testing on wikipedia.
(In reply to comment #6) > Looks like this is already fixed from my testing on wikipedia. It's not. Try putting == [[Link|<nowiki />]] == in a page and look for <span class="mw-headline" id=".7FUNIQ...QINU.7F">. However, I don't think this can be fixed at the moment, as Brion points out at bug 93 comment 30 that the parser is currently being frozen and rewritten.
(In reply to comment #7) > However, I don't think this can be fixed at the moment, as Brion points out at > bug 93 comment 30 that the parser is currently being frozen and rewritten. http://thread.gmane.org/gmane.science.linguistics.wikipedia.technical/54555
Fixed in r114231.
Looks like this is fixed. Tested with 1.21wmf8.