Last modified: 2012-06-17 19:22:28 UTC
If i am using llurl=1 on my request my parser always expect that there is an url attribute at the langlinks/ll-Tag. I think all parsers should expect this if there is not error. I would suggest that it is better to not display this langlink at all if the title contains invalid utf-8. This would be the same behaviour as done by the renderer. ~~~~
(In reply to comment #0) > I would suggest that it is better to not display this langlink at all if the > title contains invalid utf-8. This would be the same behaviour as done by the > renderer. ~~~~ As you point out, those links don't show up on the sidebar.
I'm not sure why they need to be always adding, we do a lot of things only added optionally, I don't see why the exception is needed here?
As it is, this seems like a great way to find invalid UTF-8 in the urls
When the langlinks not shown in the sidebar, why storing this item in the database? action=parse does not display the items: http://zh.wikipedia.org/w/api.php?action=parse&pageid=15669&prop=langlinks This is a page, which is not touched since a very long time (touched="2007-12-04T09:51:12Z"). After a nulledit the invalid items are gone from the database. Maybe let the api filter the invalid items out of the result?
In my opinion you should not assume, that "url" is a required param, it is optional. The module returns all entries from the database and adds extra information, when there are extra information. On an invalid title, he cannot provide extra information and that is ok. Hiding the whole output makes no sense in this case. Supporting WONTFIX.
Setting WONTFIX per comment 5