Last modified: 2014-05-30 21:11:54 UTC
When rendering a page, the result of a template with the same given parameters is only calculated the first time, and other same transclusions are taken from a cache. This is usually fine, but may be undesirable, in rare cases. For example, consider this case: lorem ipsum 1<ref>AAA</ref> {{reflist}} lorem ipsum 2<ref>BBB</ref> {{reflist}} Here, the 2nd transclusion gives a wrong result. I think it may be nice to implement a keyword, like __NOTEMPLATECACHE__ which, as its name implies, would make template re-rendered. Of course, this option wouldn't (and has not to) be used often, but it would be useful in some edge cases.
Of course, the keyword would have to be added in the template's page.
Change 99792 had a related patch set uploaded by Anomie: Add PPFrame::isVolatile and PPFrame::setVolatile https://gerrit.wikimedia.org/r/99792
Change 99793 had a related patch set uploaded by Anomie: Mark parser hook output as volatile https://gerrit.wikimedia.org/r/99793
Change 99792 merged by jenkins-bot: Add PPFrame::isVolatile and PPFrame::setVolatile https://gerrit.wikimedia.org/r/99792
Change 99793 merged by jenkins-bot: Mark parser hook output as volatile https://gerrit.wikimedia.org/r/99793
Templates containing ref or reflist tags will no longer be cached. This change will be live on all WMF wikis starting on June 12th.