On 28/05/11 04:21, Marcus Kool wrote:
Amos Jeffries wrote:
On 28/05/11 00:46, Marc Nil wrote:
<snip, snip crazy>
I also tested this option (with the sale result) I even removed the
InternetAdmin line (still the same result)
Definitely the regex bits then.
If you post the whitelist.no_limit we are able to see if there is
room for improvement.
Usually there is.
FYI he did. It was two domain names :(
As a side note for Amos: I made the URL filter ufdbGuard 70% faster
with a regex optimiser that combines N REs into one large RE composed
of the N original REs with OR logic. i.e.
RE1
RE2
...
REn
becomes
RE1|RE2|...|REn
The speed increase was measured with overall throughput so the actual
improvement for RE processing must be much higher than 70%.
I know that is one of the speed improvements Squid needs. Another is
pruning away useless "." or ".*" prefixes and suffixes.
Feel like making a patch?
Regex ACL in Squid are loaded by one very small function in
src/acl/RegexData.cc. All that has to be careful of is the -i/+i flags
which toggle case sensitivity between sets of |'able patterns.
Amos
--
Please be using
Current Stable Squid 2.7.STABLE9 or 3.1.12
Beta testers wanted for 3.2.0.7 and 3.1.12.1
