In the September 2011, December 2011, and several of the closed site data dumps the tag history is not delimited consistently across all sites. As an example from IT Security:
<row Id="48" PostHistoryTypeId="3" PostId="26" RevisionGUID="6b0fa8fb-4ae4-4d8c-8f91-d0272b10d9e0" CreationDate="2010-11-11T22:04:24.517" UserId="4" Text="ébestöpracticesàéwebsiteösecurityà" />
ébestöpracticesàéwebsiteösecurityà as the text for the list of tags.
This goes against the behaviour I had expected, namely as in the Gaming dump:
It also appears to just be the
posthistory.xml file that is affected, as the
posts.xml file contains the <> delimiters as expected.
I've taken another look at this, it seems to only affect newer sites (in both Sep & Dec dumps), which would suggest a change (charset configuration?) that was not retrospectively applied, and my incomplete download of the September dump indicates the change was made somewhere between GIS and Unix, since GIS is okay and Unix is not. (Everything else inbetween is only partially downloaded on my laptop)