Re: Real-World Impact of Hash DoS in Lua

[Florian Weimer <fw@...>]

* Roberto Ierusalimschy:

>> For 5.1, I've got a patch that replaces the string hash function with
>> the de-facto industry standard (one of Bob Jenkin's functions).  I
>> could tidy up the patch and post it if there is enough interest.

> What were the gains?

The sensitive to input with regular patterns was gone, but hashing was
a bit slower.  Some microbenchmarks were slower, some were faster.  My
test case for collisions did not work anymore.