cmdTable sentinel is incorrect about redis HOT 8 CLOSED

Thanks! fixing this in SVN, I'm writing a regression test for the issue. I'm 
curious,
did you spot it because of a crash on unknown command or looking at the source 
code? :)

It crashed when running the "Non existing command" test from the testsuite.

ah ok, pretty strange that valgrind didn't reported this even running the full 
'make
test' under valgrind, so it didn't fixed before. Btw now it's fixed under SVN. 
Thanks
for the help. I'll probably release beta-9 tomorrow with just this fix since 
it's a
bit critical if the server can crash with the standard test.

I mean beta-8 :) not 9

... which is pretty much the same as the test you just added.

I bet the original test worked for you by fluke because whatever happened to be 
in
memory past the end of the array happened to be zeroes.

Oh weird that valgrind didn't catch it.

Yes sorry I removed the duplicated test, I added it by instinct since it was 
out of
my mental scheme that there was already a test about it and still the bug was 
there.
Now I removed the new test since it's absolutely useless. I'll close the bug 
only
after beta-7 will be deprecated and beta-8 released. Thanks again.