fre 2010-06-11 klockan 22:35 +1200 skrev Amos Jeffries:
Was there anything to be added before this gets to 3.1?
comm loops using fd_set need to limit to FD_SETSIZE or there will be
memory corruption and other badness if someone tries to configure a high
limit.
Regards
Henrik
ons 2010-06-09 klockan 10:34 +0200 skrev Kinkie:
You're probably missing Berkeley db version 1.8.5 , but have a more
recent version (and incompatible), and this confuses autoconf.
IIRC the autoconf-refactor branch has some better detection logics,
but it hasn't been merged to trunk or to 3.1
tis 2010-06-08 klockan 23:15 +0200 skrev Christian:
+DEFAULT_SWAPDIR=$localstatedir/cache
Looks fine, but the above should be
DEFAULT_SWAPDIR='$(localstatedir)/cache'
Regards
Henrik
ons 2010-06-09 klockan 09:42 -0500 skrev Dean Weimer:
Making all in session
gcc -DHAVE_CONFIG_H -I../../.. -I../../../include -I../../../src
-I../../../include -I. -I/usr/local/include -Wall -Wpointer-arith
-Wwrite-strings -Wmissing-prototypes -Wmissing-declarations -Wcomments
tis 2010-06-08 klockan 11:13 -0500 skrev Dean Weimer:
As of 3.1.4 this compiles fine on FreeBSD 8.0, and 7.1. However, the FreeBSD
7.2 system I have still has this problem with 3.1.4.
Odd.
What error do you get?
Regards
Henrik
tis 2010-06-08 klockan 23:30 +1200 skrev Amos Jeffries:
Says something sad about our present audit process for nobody to have
noticed that in 4 days :(
Well, can't comment much on the code before it's in the tree and hits
the compiler, and trust you to shuffle things reasonably to the right
tis 2010-06-08 klockan 02:41 -0600 skrev Amos Jeffries:
if test $HOST = squid-cache.org ; then
- MD5=python /usr/local/share/python2.4/Tools/scripts/md5sum.py -
+ MD5=python /usr/local/share/python2.6/Tools/scripts/md5sum.py -
Gah.. hardcoded paths again.
Was there some reason why
fre 2010-06-04 klockan 17:40 +0100 skrev Markus Moeller:
That is a bit my problem. For example if I use heimdal kerberos libraries
version 1.3.1 I need
-lgssapi -lheimntlm -lkrb5 -lhx509 -lcom_err -lcrypto -lasn1 -lwind -lroken
-lcrypt -ldl -lresolv -pthread
for older 0.7.2 I need
mån 2010-06-07 klockan 10:49 +0200 skrev Kinkie:
Sure. The downside I see is that then any host wishing to bootstrap
needs to have pkgconfig available. One more dependency to account for.
We can provide fallback methods for the most important parts if needed.
Regards
Henrik
mån 2010-06-07 klockan 10:38 +0200 skrev Kinkie:
Well, what I recall from the discussion is that valgrind support for
squid acts by changing the memory allocation mechanisms to be
valgrind-friendly. But I'm okay with any decision :)
It does change the cbdata implementation to the older cbdata
fre 2010-06-04 klockan 01:03 +0200 skrev Kinkie:
Those kinds of checks ae being phased out of the configure.in script,
unless they are hidden by other macros. In general using main to test
for lib existence is IIRC not recommended by autoconf docs..
Using main doesn't work properly, and does
fre 2010-06-04 klockan 00:19 +0100 skrev Markus Moeller:
What is the recommended way of doing it as I am adding a module which
requires library checks and I don't know always what function is best to
check on ?
A function you know is implemented by the named library.
Note: watch out for
tor 2010-06-03 klockan 00:00 +1200 skrev Amos Jeffries:
Tsantilas Christos wrote:
Hi all,
This patch add logging of the local TCP port used by transactions with
http servers
The new log format code is %lp
In the case there are several server-side connections logs the port of
lör 2010-05-29 klockan 18:58 +1200 skrev Amos Jeffries:
+CXXFLAGS=$CXXFLAGS -nostdlib
+AC_SEARCH_LIBS([malloc],[tcmalloc jemalloc dlmalloc ptmalloc3
nedmalloc ptmalloc2 malloc])
+CXXFLAGS=$saved_flags
I am not very comfortable with this. -nostdlib is a GCC option, and also
there is no
lör 2010-05-29 klockan 15:06 +1200 skrev Amos Jeffries:
This is the bit intended for 3.2+ right?
Should probably go back to earlier releases as well. Loss of mempool
statistics is highly undesired.
In addition the small change in trunk is missing configure.in help text
correction. Thought I
lör 2010-05-29 klockan 22:36 +1200 skrev Amos Jeffries:
That single squid_curtime dependency is exactly the remaining problem
with the non-squid binaries in AIX and Cygwin builds. Despite the
binaries that are breaking making no use of this memory code.
That dependency never brings in the
lör 2010-05-29 klockan 15:17 +1200 skrev Amos Jeffries:
Also needs release notes update for added option
--enable-chunkedmempools and removed option --enable-mempools.
I think I'll actually drop the configure parameter for now. That code is
now experimental, and can be enabled by a
lör 2010-05-29 klockan 22:36 +1200 skrev Amos Jeffries:
Okay. Forget the shuffling in this commit then.
+1.
Merged.
Regards
Henrik
fre 2010-05-28 klockan 17:08 +1200 skrev Amos Jeffries:
Right now in light of the issues that keep cropping up my vote for
squid-3.1 is for (a). Dropping of the impossible settings in squid.conf,
with real default-disabled mempools. Plus some clear documentation of
the environment variable
lör 2010-05-29 klockan 01:59 +1200 skrev Amos Jeffries:
Hmm, being drop-in. Then this would be all that is needed adding to
configure.in yes?
AC_SEARCH_LIBS([malloc],[tcmalloc gnumalloc malloc])
Correct.
We may also want to add some support for mgr:mem page and stuff, but
that's extra
fre 2010-05-28 klockan 11:21 +0200 skrev Henrik Nordström:
Not what I meant by 'a', but obviously also is an alternative
c) Change the pools to default fully off and hidden away, requiring an
environment variable to be set to enable.
Unfortunately this seems to also disable the mempools
This change splits MemPool.cc into framework and specific allocators
MemPool.cc general framework
MemPoolChunked.cc chunked allocator
MemPoolMalloc.cc malloc allocator, no freelist (yet)
This may also fix some statistics bugs. There was some oddness there
with a bit of confusion between
tor 2010-05-27 klockan 13:16 +1200 skrev Amos Jeffries:
Auto-enable is great for encouraging feature use in self-builds, but for
distros its just added work for the downstream maintainers.
Well, things like IPv6 is auto-enable in most packages now, not just
Squid.
But I have expressed this
fre 2010-05-28 klockan 00:34 +1200 skrev Amos Jeffries:
vanligt textdokument-bilaga (r10511.diff)
=== modified file 'helpers/basic_auth/DB/basic_db_auth.in'
--- a/helpers/basic_auth/DB/basic_db_auth.in 2010-04-26 21:59:49 +
It has come to light that the chunked allocator have a couple of
fundamental design issues.
The primary is that free space management becomes overly complex,
causing significant performance issues.
The second is that you can not really disable the use of pools by
configuration, even if we claim
ons 2010-05-26 klockan 21:43 +1200 skrev Amos Jeffries:
I was hoping that after your autoconf rework it would become a simple
matter of making the initial default setting for most option tests a
simple variable assignment of the value (auto or no) as set by this option.
Not sure it matters.
lör 2010-05-22 klockan 10:57 -0600 skrev Alex Rousskov:
That worked. I have at least 2.1.1 everywhere now.
Cool.
Seems all of us is now at bzr 2.x, which means we can upgrade the main
repository format to 2a. Good, did not really expect this to happen for
another 6 months.
Anyone who have
ons 2010-05-19 klockan 12:14 +1000 skrev Mark Nottingham:
which to me says that it's simply an indication of whether it's in-cache or
not, not whether it's stale. Since the response headers come back on the HTCP
response, the querying peer can figure out whether or not it's fresh --
sön 2010-05-23 klockan 15:23 +0200 skrev Kinkie:
Do the two lookups by some reason need to be in separate queries?
Can't we just ask for the A and in the same query packet?
Sorry for my ignorance..
I wonder how many resolvers will actually handle both queries in the
same
sön 2010-05-23 klockan 16:30 -0600 skrev Alex Rousskov:
bzr diff -r ancestor:. --new lp:~yadi/squid/cleanup-comm
I used this trick for the review. For the record, the diff includes
reversal of recent trunk changes but it is probably possible to avoid
that if the last branch merging point
mån 2010-05-24 klockan 08:46 +0200 skrev Henrik Nordström:
a bit confusing command this..
And just found another easier, optimized and lightweight way of doing
the same.
bzr merge --preview lp:~yadi/squid/cleanup-comm
Regards
Henrik
lör 2010-05-22 klockan 17:04 -0600 skrev Alex Rousskov:
Is there an RFC or somesuch that recommends parallel lookups over
sequential ones or vice versa?
IIRC the RFC recommends lookup and only if that returns an empty
respose then try A.
But at least Linux does the lookup in parallel
sön 2010-05-23 klockan 23:55 +1200 skrev Amos Jeffries:
Hno; EDNS fields were trivial to add as well and close the EDNS bug. I
have some experimental code on west adding them which I can push your
way if you like. Just stupid build errors AFAIK.
Is that for the larger query size support? Or
sön 2010-05-23 klockan 13:25 +0200 skrev Kinkie:
Do the two lookups by some reason need to be in separate queries?
Can't we just ask for the A and in the same query packet?
Sorry for my ignorance..
I wonder how many resolvers will actually handle both queries in the
same request...
sön 2010-05-23 klockan 15:23 +0200 skrev Kinkie:
No idea. I guess it depends on how magic is..
* Practically no software ever sends DNS requests with more than one
query.
* Several DNS servers are known to fail with server failure on an
query on a known label even when specifications
fre 2010-05-21 klockan 20:52 -0600 skrev Alex Rousskov:
Pushed to launchpad: lp:~yadi/squid/cleanup-comm
How can I review the changes as one patch without checking out your branch?
Make sure you have an up to date copy of trunk, then from that run
bzr diff -r ancestor:. --new
fre 2010-05-21 klockan 20:39 -0600 skrev Alex Rousskov:
On 05/19/2010 06:25 PM, Henrik Nordström wrote:
For repository maintenance reasons I need to know which minimum bzr
version all who work with the Squid repository need to be able to use.
And I mean all, including those who do not have
This was intentionally not wrapped in an ifdef as the code here is not
IPv6 dependent. If anything it's actually IPv4 dependent...
lör 2010-05-22 klockan 17:59 +1200 skrev Amos Jeffries:
+#if USE_IPV6
/* Handle IPv6 over IPv4-only socket case.
* this case must presently be
Skickat: 21 maj 2010 05:43 -07:00
Till: Henrik Nordström hen...@henriknordstrom.net, Alex Rousskov
rouss...@measurement-factory.com, Squid Developers
squid-dev@squid-cache.org
Ämne: Re: autoconf-refactor and netfilter-based transparent proxy
Hi all.
To avoid risking this to happen again, i'll do
fre 2010-05-21 klockan 09:37 -0600 skrev Alex Rousskov:
I do not really know what you mean. If you are comfortable with your
changes, post your changes for review, and nobody objects to a commit,
you should commit your changes to trunk.
In this case I am happy to have Kinkie review his own
tis 2010-05-18 klockan 23:34 + skrev Amos Jeffries:
I've discovered the VC connections in DNS will need a re-working to handle
the new TCP connection setup handling. I've left that for now since it
appears that you are working on redesigning that area anyway. The new setup
routines will
tis 2010-05-18 klockan 23:42 + skrev Amos Jeffries:
I have the impression the purge tool is probably worth bundling with Squid
alongside squidclient and cachemgr.cgi in the tools/ section.
Before the license was incompatible. Haven't looked if the license
changed at the end of the project.
ons 2010-05-19 klockan 22:46 +1200 skrev Amos Jeffries:
Looks as compatibile as we can get.
Agreed.
Regards
Henrik
ons 2010-05-19 klockan 10:03 -0500 skrev senad.ci...@thomsonreuters.com:
Not sure what went wrong, but since changes in each of the 3 files are
minimal I included differences as attachments (new version for each
file is on top, old version is on bottom):
Probably got converted from UNIX to DOS
tis 2010-05-18 klockan 16:25 -0500 skrev senad.ci...@thomsonreuters.com:
Attached file is generated patch for the issue where squid does not
contact its siblings for the stale resources. New directive has been
added to make this behavior optional.
Corrected patch attached showing only your
For repository maintenance reasons I need to know which minimum bzr
version all who work with the Squid repository need to be able to use.
And I mean all, including those who do not have direct commit access.
I.e the output of bzr --version | head -1 on the oldest platform you
need or want to
tor 2010-05-20 klockan 12:00 +1000 skrev Mark Nottingham:
H, is setting the default to on a bad thing?
Probably not, as the code takes care of allow-miss.
related to this the behavior when the sibling rejects the request need
to be verified. This change will make only-if-cached rejections in
tis 2010-05-18 klockan 15:12 +1000 skrev Mark Nottingham:
/*
* The 'need_validation' flag is used to prevent forwarding
* loops between siblings. If our copy of the object is stale,
* then we should probably only use parents for the validation
* request.
tis 2010-05-18 klockan 20:49 +1200 skrev Amos Jeffries:
Nevermind sorry. I was overloooking the #ifdef IPV6_V6ONLY in
comm_set_v6only().
The debugs() in that #else case should be bumped down to something less
annoying now. It _will_ occur on each socket open on WindowsXP and maybe
others.
tis 2010-05-18 klockan 20:54 +1200 skrev Amos Jeffries:
These fixes are not really undoing the bug 222 fix are they? mroe
replacing it with a better fix AIUI.
Can we close Bug again now?
I suppose so.
But need to fix DNS before I am happy.. currently the A lookup is
heavily penalized.
tis 2010-05-18 klockan 22:59 +1000 skrev Mark Nottingham:
Any thoughts about moving where refreshCheck is called for HTCP?
Haen't looked at it. What is the problem?
Regards
Henrik
sön 2010-05-16 klockan 11:58 +1200 skrev Amos Jeffries:
By existence I meant check if it is defined.
If the macro is not defined it's an --enable-ipv6 fail, same as if
PF_INET6 was not defined.
Why? It's not critical for IPv6.
Regards
Henrik
lör 2010-05-15 klockan 17:56 +1200 skrev Amos Jeffries:
But that leaves us without support for v6 on MacOSX and OpenBSD, and
WindowsXP.
Accepting requests should work already in 3.1, but user need to
configure two http_port one for ipv6 and one for ipv4.
But yes, forwarding would not work
lör 2010-05-15 klockan 10:42 +0200 skrev Henrik Nordström:
But I honestly do not see why we need to configure run test for any of
this, other than that we can compile. A Squid compiled with IPv6 support
will now run just fine even if the IPv6 stack is not there at runtime,
and the only use
lör 2010-05-15 klockan 22:47 +1200 skrev Amos Jeffries:
I think we should retain a test for existence of IPV6_V6ONLY though. It
can be a if-compiles and will catch the WindowsXP etc cases.
Why? If IPV6_ONLY is there then the code will always request it to be
disabled.
The .cc/.h changes
And corrected version.. was an unrelated configure.in change which did
not belong here.
lör 2010-05-15 klockan 23:22 +0200 skrev Henrik Nordström:
lör 2010-05-15 klockan 22:47 +1200 skrev Amos Jeffries:
I think we should retain a test for existence of IPV6_V6ONLY though. It
can
fre 2010-05-14 klockan 18:11 +1200 skrev Amos Jeffries:
I can't seem to get tcp_outgoing_address to failover cleanly or mark the
v6 address bad with this one when there is no v6 link on the box.
If you have no IPv6 stack at all then it will query for records,
but silently skip them.
It
fre 2010-05-14 klockan 18:11 +1200 skrev Amos Jeffries:
I can't seem to get tcp_outgoing_address to failover cleanly or mark the
v6 address bad with this one when there is no v6 link on the box.
What did you have tcp_outgoing_address set to? And what kind of failover
are you expecting?
This
fre 2010-05-14 klockan 18:11 +1200 skrev Amos Jeffries:
Henrik Nordström wrote:
fre 2010-05-14 klockan 16:47 +1200 skrev Robert Collins:
+1
Applied.
I can't seem to get tcp_outgoing_address to failover cleanly or mark the
v6 address bad with this one when there is no v6 link
fre 2010-05-14 klockan 18:20 +1200 skrev Amos Jeffries:
The problem is that the flag IPV6_V6ONLY is a standard define. It's
present on all IPv6 systems. The only way to identify the stack
operation, is to use it and see what error comes back.
Is there stacks having this define but not
As you may have noticed I have been looking at the IPv6 support in the
last days, and now I am looking at how we perform DNS queries.
Today we first do an query, followed by an A query. As we anyway do
both queries it would be better if we ran both in parallell. The
benefits of this is
for the A lookup to succeed is close to nil.
But the while retry mechanism is broken by design when it comes to
active errors and breaks down if a resolver nameserver is
malfunctioning.
fre 2010-05-14 klockan 21:38 +0200 skrev Henrik Nordström:
As you may have noticed I have been looking at the IPv6
Take two.
The change is that this version also tries to skip IPv4 addresses if the
socket is IPv6-only. This case should have been handled by existing code
paths, but the way that was done was quite twisted and apparently broke
down by my changes to commResetFD.
This patch backs out part of the
fre 2010-05-14 klockan 11:23 +1200 skrev Amos Jeffries:
It does not mark IPv6 'BAD' in the case where the link is down or boudn
wrong. This is fine since the failover works, but we may want to look
into it later.
Not 100% sure what you try to say here
The following is intentional however:
Take three
Differences from take two is that
* the test for IPv6 over IPv4 have been corrected, was triggering
a bit too much, blocking IPv6 access if tcp_outgoing_address not
set
* Set errno to ENETUNREACH when destination blocked by family, to
match what
fre 2010-05-14 klockan 03:46 +0200 skrev Henrik Nordström:
Take three
merged.
The attached patch makes squid automatically fall back on IPv4 operation
if it fails creating an IPv6 socket. This may happen if Squid is built
with IPv6 support enabled but no IPv6 stack is available when it runs.
# Bazaar merge directive format 2 (Bazaar 0.90)
# revision_id:
fre 2010-05-14 klockan 07:09 +0200 skrev Henrik Nordström:
The configure test for ipv6 split-stack configuration fails if there is
IPv6 headers but the build host do not have IPv6 enabled.
Correction, it's the v4mapped test that fails.
Please try to figure out a compile-only test
fre 2010-05-14 klockan 16:47 +1200 skrev Robert Collins:
+1
Applied.
ons 2010-05-12 klockan 15:07 +1200 skrev Amos Jeffries:
http_access allow all
acl to_ipv6 dst ipv6
tcp_outgoing_address 2001:... to_ipv6
tcp_outgoing_address 192.168 !to_ipv6
I would guess this fails if the address Squid then tries to connect to
ends up being ipv4, in similar manners
tis 2010-05-11 klockan 23:14 +1200 skrev Amos Jeffries:
The v6 worry was that it might have broken v4/v6 gateways. see Bug
comment #3. This no longer seems to be the case for
non-tcp_outgoing_address configurations.
mixed v4/v6 usage with tcp_outgoing_address produces a constant
This patch backs out part of the patch for Bug # and replaces it by
crudely cycling over the available addresses, trying to skip over
addresses not compatible with the current socket.
This solves issues seen when using tproxy or tcp_outgoing_address and
DNS of the requested host returns
I propose to drop the DNS_CNAME code from the tree. Recursing CNAME
records is not something we ever should be doing as it's the job of the
queried resolver, and having that code in tree but always disabled adds
no value only overhead.
See earlier discussion from when the code was added and then
mån 2010-05-03 klockan 21:59 +1200 skrev Amos Jeffries:
So an automatically generated outgoing IP address based on an iface name
then?
Selecting the iface, then picking from one of its addresses?
Sounds like it should actually work.
Donẗ forget to add whildcard interface name syntax
mån 2010-05-03 klockan 15:24 -0430 skrev Robert Marcano:
I understand, but we are using tcp_outgoing_address because it supports
ACLs and we route some users to a dedicated ISP exclusive to them.
Perfectly valid reason.
It breaks server side persistent connections so we disable it, but I
I saw that the change to announce HTTP/1.1 to clients has sneaked in in
3.1.2. Is this really intentional? Can't remember seeing any discussion
about this.
I know we discussed announcing HTTP/1.1 towards servers, but imho we are
not yet ready for announcing HTTP/1.1 towards clients.
- Not yet
lör 2010-05-01 klockan 08:55 -0600 skrev Alex Rousskov:
Chunked requests: The current buffer-and-forward code is probably
compliant. It is also inefficient and limits the size of the request,
but I doubt that violates the standard.
buffer forward with content-length in itself do not violate
lör 2010-05-01 klockan 21:28 +0200 skrev Kinkie:
It takes two days to run a build, so it won't be auto-triggered often,
and we can probably afford to follow only one branch. But it works :)
Cool!
A suggestion to speed things up considerably is to use distcc with a
cross-compiler, allowing the
sön 2010-05-02 klockan 14:31 +1200 skrev Amos Jeffries:
In particular do you mean requests? as in chunked POST/PUT bodies.
or responses to a request accepting chunks?
requests with a request entity sent using chunked encoding.
- Expect handling not in place
You were the one who
tor 2010-04-29 klockan 15:53 -0430 skrev Robert Marcano:
I am attaching a patch for 3.0.x versions (I will forward port if there
is interest and change recommendations). This patch adds a new option to
Squid that allows it to forward the current authenticated user to the
next proxy (or the
Vidarebefordrat meddelande
Från: Jan Rheinländer jrheinlaen...@gmx.de
Reply-to: resap-...@iamafg.org
Till: squid-b...@squid-cache.org
Ämne: clientside_tos and qos_flows
Datum: Tue, 27 Apr 2010 10:23:46 +0430
Hi,
I tried the following:
clientside_tos 0x30 update
qos_flows
mån 2010-04-26 klockan 00:11 + skrev Amos Jeffries:
Is is actually safe/good to default-enable the mem-gen-trace?
Yes, if the needed requirements are found.
The only risk is that since Squid performs more stuff at fatal errors
which may fail if Squid is crashing due to a SIGSEGV. This
tis 2010-04-27 klockan 00:27 +0200 skrev Henrik Nordström:
mån 2010-04-26 klockan 00:11 + skrev Amos Jeffries:
Is is actually safe/good to default-enable the mem-gen-trace?
Yes, if the needed requirements are found.
Hmm.. no.
automatic stack traces is safe to auto-enable
ons 2010-04-07 klockan 20:27 +0100 skrev Markus Moeller:
Would it make sense to define in squid two new configuration options to
control Negotiate authentication ? I am thinking of adding
Negotiate-NTLM
and
Negotiate-Kerberos
I would prefer a wrapper helper doing this selection.
tis 2010-03-30 klockan 18:15 +1000 skrev HC Barfield:
hi,
i was wondering if you can release squid with the 'intermittant connections
patch' so that it is more suitable for home users and so we got some good
FREE software, as the alternatives are not convienent.
Found that patch in the
tor 2010-03-25 klockan 18:01 +0100 skrev Kinkie:
He didn't threaten any, but I didn't specifically ask for promises :)
The worst I can think of is that it will mandate all developers to
have halfway recent bzr installs.
2a will basically require bzr 1.18 or later to be useful iirc.
Upgrading
lör 2010-03-20 klockan 03:50 +1300 skrev Amos Jeffries:
=== modified file 'test-suite/test_tools.cc'
--- a/test-suite/test_tools.cc2009-08-28 01:44:26 +
+++ b/test-suite/test_tools.cc2010-03-19 14:50:48 +
@@ -5,7 +5,6 @@
// XXX: This file is made of large pieces of
lör 2010-03-20 klockan 23:04 +1300 skrev Amos Jeffries:
The test that was failing for the last day or so was doing so because it
linked SquidNew.o + test_tools.o.
Yes, which was one of the items in src even, not a test.
It seems to pass all the testbed tests, so I assume the other ones that
ons 2010-03-17 klockan 03:26 + skrev Amos Jeffries:
Since we are working on this. Is new/delete operators not something that
should be in src/base libbase anyway?
Inline operator overloading is a little tricky to have in a lib
unfortunately, due to the same reasons ufsdump linking failed.
tis 2010-03-16 klockan 01:15 +0100 skrev n...@squid-cache.org:
/bin/sh ../libtool --tag=CXX --mode=link g++ -Wall -Wpointer-arith
-Wwrite-strings -Wcomments -Werror -g -pipe -g -o squid AclRegs.o
AuthReg.o AsyncEngine.o cache_cf.o ProtoPort.o CacheDigest.o cache_manager.o
carp.o
mån 2010-03-15 klockan 20:28 +0100 skrev Luigi Gangitano:
I've got the same on Debian unstable with 3.1.0.17 and 3.1.0.18.
g++ -DHAVE_CONFIG_H -DDEFAULT_CONFIG_FILE=\/etc/squid3/squid.conf\
-DDEFAULT_SQUID_DATA_DIR=\/usr/share/squid3\
-DDEFAULT_SQUID_CONFIG_DIR=\/etc/squid3\ -I..
tis 2010-03-16 klockan 00:49 +0100 skrev Henrik Nordström:
/* #define _SQUID_EXTERNNEW_ */ /* Why? kinkie */
That why is explained in include/SquidNew.h and the fact that this
define is now commented out may well explain thing a bit.. Plus the fact
that ufsdump do not use SquidNew.cc which
While investigating why digest auth wrongly indicated stale=false on
unknown nonces even when all the logics for fixing that has been
forwardported since way back I stumbled across this little difference
between Squid-2 Squid-3 in how they fix up auth headers on failed auth
requests. In Squid-2
mån 2010-03-01 klockan 14:46 -0700 skrev Alex Rousskov:
in hope that it would be useful for Squids that default to HTTP/1.0 and
those that default to HTTP/1.1. I am not sure we will ever need to
downgrade to HTTP/1.0 but it is certainly possible, especially if we
default to HTTP/1.1 (is not
tis 2010-03-02 klockan 11:17 +1300 skrev Amos Jeffries:
* stripping Expect: headers on requests (by skipping 417 abort plus
request_header_access Expect deny all)
Haven't seen any need for skipping/filtering the Expect header. Squid-2
forwards any received Expect headers.
* ignoring
tis 2010-03-02 klockan 00:06 +0100 skrev Henrik Nordström:
mån 2010-03-01 klockan 14:46 -0700 skrev Alex Rousskov:
in hope that it would be useful for Squids that default to HTTP/1.0 and
those that default to HTTP/1.1. I am not sure we will ever need to
downgrade to HTTP/1.0
tis 2010-03-02 klockan 12:34 +1300 skrev Amos Jeffries:
The only risk I can really see is with compliant servers. If they see us
as 1.1 relaying an Expect: there is a reasonably high risk they might
actually try to use it. As they should.
Which won't hurt compared to today assuming we are
tis 2010-03-02 klockan 01:07 +0100 skrev Henrik Nordström:
tis 2010-03-02 klockan 12:34 +1300 skrev Amos Jeffries:
The only risk I can really see is with compliant servers. If they see us
as 1.1 relaying an Expect: there is a reasonably high risk they might
actually try to use
tis 2010-03-02 klockan 01:14 +0100 skrev Henrik Nordström:
Which won't hurt compared to today assuming we are swallowing the 100
responses, which we already do due to broken servers sending them to us
even in HTTP/1.0..
Or.. are we.. can'ẗ find the code for that in Squid-3. Have a memory
The main server has now been moved as announced. All services should be
up by now, but if you find something not working please speak up.
Regards
Alex, Duane Henrik
401 - 500 of 597 matches
Mail list logo
