Stuff Michael Meeks is doing

Fixed a mind-blowingly long standing evil ~/.recently-used XML file corruption bug in OO.o - horay for Kelli actually filing the bug.
Poked at acroread performance: turns out you can run it quite nicely with: time acroread -toPostScript /demo/a.pdf /tmp/a.ps a couple of times then do echo "3" > /proc/sys/vm/drop_caches, look for the amazing vanishing time:
```
time acroread -toPostScript /demo/a.pdf /tmp/a.ps:
        1st time:    2nd time:
real    0m32.221s    0m1.586s
user    0m1.684s     0m1.108s
sys     0m1.080s     0m0.464s
```
explaining the gap between 2.6 seconds and 32 seconds is (perhaps) the idea of reading every font on the system, twice (in 2 separate passes) - the strace -ttt shows 32 seconds for that. Gratifying to see that other people have sillier problems than OO.o, will try to poke them: surely .pcf.gz fonts aren't worth playing with anymore(?).
My (amazing) script and C file reduce this to 10 secs: with some more path tweaking no doubt getting more would be easy enough.
Played with the Google web-spreadsheet toy, blown away by the lack of substance. Hacked up a perl script, and generated some more reliable numbers on our OO.o up-streaming progress. We seem to have peaked just under 1000 patches, unfortunately, as fast as we manage to nudge some bits up-stream, we generate more.