Autocorrect doesn’t know math

The autocorrect feature in Gmail has the unfortunate but hilarious habit of vigorously changing standard math terms into free-associative nonsense.  Here are some highlights (and I might add to this list from time to time):

• “Igusa varieties” -> “Iguana varieties”
• “Zariski topology” -> “Czarist topology”
• “Gelfand spectrum” -> “Gelatin spectrum”
• “cokernel” -> “cockerel” (my favorite so far)

Anyway, the paper should basically be stable at this point, with the exception that $\S4.3$ will probably be rewritten to some degree once Peter’s six-functors book is done.  The only real difference from the the first version is that the material around the “pointwise criterion” in $\S2.2$ has been streamlined and clarified a bit.  All comments, questions or corrections are very welcome!