# Duplicate banner breaks LaTeX

On several recent occasions, after closing a question as a possible duplicate the automatic banner breaks the $\LaTeX$ requiring manual intervening.

I don't recall it was like that before, and I suppose that something was changed in the way the system fetches the question's title. It does not break in the announcing comment automatically added when first suggesting a duplicate, though.

One recent example is Vector space over $F_p$ not the union of $k\leq p$ subspaces, but I recall doing that at least once more in the past week.

• It always was like that since I'm here, it was brought up a long time ago in this thread. I don't quite understand the fix that was made but obviously it doesn't really work or at least it doesn't handle all exceptions.
– t.b.
May 25, 2012 at 11:13
• @t.b.: Interesting. Well, I guess that [status-completed] is not really completed after all. (Perhaps a recent change broke the fix?)
– Asaf Karagila Mod
May 25, 2012 at 11:14
• No, I don't think so. I seem to remember to have fixed broken LaTeX in duplicate links during the entire time I'm here.
– t.b.
May 25, 2012 at 11:15
• @t.b.: Your first comment here is from two minutes, that's not a long time. :-P (I don't remember fixing duplicate banners before last week and I don't remember anyone else fixing... but my memory is huge holes in it.)
– Asaf Karagila Mod
May 25, 2012 at 11:17
• Here's an example of an older post where a similar thing happened as in the example you bring up here. Just in case it is related: here's another meta-question where problems with inequality signs were brought up.
– t.b.
May 25, 2012 at 11:33
• @t.b.: I see. Then I suppose either the developers see that the problem persists, or once again we'll have a "next deployment..." and [status-completed].
– Asaf Karagila Mod
May 25, 2012 at 11:37

This one involves escaping already-escaped HTML entities. Greater-than and less-than symbols are normally escaped automatically when the title is formatted for the duplicate link. When processing MathJax sections, ampersands are escaped. So > becomes &amp;gt; in the final output, which doesn't do anything good.