On 5 December 2012 05:19, Tim Starling <[email protected]> wrote:

> The jobs were probably not parsing Pagina Principale, they were
> probably parsing some page with an infinite loop of #lst invocations.
> Sorry, I didn't get a backtrace.
>

On 6 December 2012 00:51, Platonides <[email protected]> wrote:

> This is obvious, but if the "new lst" allows infinite loops, it MUST NOT
> be deployed. Not even if it rendered the Pagina Principale acceptably.


I actually interpreted Tim's mail as a metaphorical 'infinite loop' -
something nested very deeply which takes a long time - not as a real
infinite loop (especially as that will just exhaust the stack and . Just to
be clear about this: there is a mechanism that prevents recursive
transclusions, and the page parses on test2wiki, so I'm guessing the origin
was not a real infinite loop - just too much computational effort.

On the code deployment side of things - Aaron had already reverted all
changes since the last known working revision. I have cherry-picked all
non-parser changes that were committed after that point in time, and most
of these have now been merged again (thanks for those improvements, Brad,
Reedy & tpt!).

I also just realized there is no link to the actual commit in this thread -
please see
https://gerrit.wikimedia.org/r/gitweb?p=mediawiki/extensions/LabeledSectionTransclusion.git;a=commitdiff;h=340dc534bb3e9adbd89c27702afa6dc5a5d7ca6afor
that.

Best,
Merlijn
_______________________________________________
Wikitech-l mailing list
[email protected]
https://lists.wikimedia.org/mailman/listinfo/wikitech-l

Reply via email to