Henrik Nordstrom wrote:
On ons, 2008-08-13 at 16:25 +1200, Amos Jeffries wrote:
Or maybe = "" should be equivalent to SetEmpty() in which case the
original code works, and can be simplified a lot... not sure how
IPAddress is supposed to work here, if it at all is supposed to be used
in = ""..
It should be capable of accepting it. "" == NULL == "garbage" to IPAddress.
Assigning "garbage" is an error.
What should the IPAddress be set to in such case?
IIRC, the ACLIP::match() logics depend on addr2 being either the end of a
range different to addr1 or empty for optimal testing. addr1==addr2 is
worst-case match() state for every request.
I suspect normalising all specifications into ranges may actually be the
optimal case.. that way the match function becomes very simple, just a
memcmp if you like.. (well, two memcmp, one for each bound, and
preferably with alignment attributes so GCC optimizes them fully). But
it requires finally dropping support for non-contiguous netmasks.
Regards
Henrik
Bah. I've just had a brain re-adjustment.
NO_ADDR == 255.255.255.255 == invalid parse request, as per standard
netinet/* definitions. Please disregard what I said in the last few
emails about it being == empty.
The NO_ADDR is used because OS are supposed to throw errors if anything
tries to use it. So we can track if squid is attempting to use a bad
address. If its set to empty or no-addr it would syntactically mean a
badly parsed address would be convertible to the default host IP by
squid (outbound send to/from ANY_ADDR may become broadcast or unicast
local host) and possibly used with unknown host kernel behavior.
Amos
--
Please use Squid 2.7.STABLE3 or 3.0.STABLE8
