]> git.ipfire.org Git - thirdparty/squid.git/log
thirdparty/squid.git
24 years agoAdded me as an official maintainer of this helper.
hno [Mon, 16 Apr 2001 03:05:49 +0000 (03:05 +0000)] 
Added me as an official maintainer of this helper.

24 years agoAbility to search for the user DN to log in as
hno [Mon, 16 Apr 2001 03:04:30 +0000 (03:04 +0000)] 
Ability to search for the user DN to log in as

Command line options for all configuration data. No longer any need
to edit squid_ldap_auth.c, just specify the correct options from
squid.conf.

Arguments:
   -b basedn
   -s searchscope
   -f searchfilter
   ldap_server_name

More documentation will be written later, explaining how to use the
searchfilter and scope, with examples on how to use this to log in
to Microsoft active directory.

24 years agoSSL->HTTP gatewaying support by Benno Rice
hno [Sun, 15 Apr 2001 00:23:41 +0000 (00:23 +0000)] 
SSL->HTTP gatewaying support by Benno Rice

Allows Squid in accelerator mode to listen for https requests
as well as http requests.

(ssl_support.[ch] was forgotten in the big commit)

24 years agoRan autoconf, autoheader
hno [Sat, 14 Apr 2001 19:11:28 +0000 (19:11 +0000)] 
Ran autoconf, autoheader

24 years agoBasic support for Vary:ing objects.
hno [Sat, 14 Apr 2001 06:25:17 +0000 (06:25 +0000)] 
Basic support for Vary:ing objects.

When an URL is found to Vary a dummy placeholder object is created
to hold the Vary header information, and the store key algorithm is
changed to include the indicated headers.

24 years agoSSL->HTTP gatewaying support by Benno Rice
hno [Sat, 14 Apr 2001 06:03:19 +0000 (06:03 +0000)] 
SSL->HTTP gatewaying support by Benno Rice

Allows Squid in accelerator mode to listen for https requests
as well as http requests.

24 years agoAgain, fix bandwidth spikes seen when the origin server did not process
hno [Fri, 13 Apr 2001 20:59:11 +0000 (20:59 +0000)] 
Again, fix bandwidth spikes seen when the origin server did not process
a range request.

Ideally the whole "we_do_ranges" check should only be done once, and replies
should only be range processed if this is true. Currently we have the check
in two, maybe even three places is seems.

24 years agoFrom Andres Kroonmaa:
hno [Tue, 10 Apr 2001 23:01:35 +0000 (23:01 +0000)] 
From Andres Kroonmaa:
avoid needless malloc/free of 2 byte string

24 years agoFrom Andreas Kroonmaa:
hno [Tue, 10 Apr 2001 23:00:50 +0000 (23:00 +0000)] 
From Andreas Kroonmaa:
bug in util.c caused xmalloc_stats to count xcalloc stats wrong

24 years agoDigestFetchState cbdata type moved to peer_digest.c
hno [Tue, 10 Apr 2001 04:26:00 +0000 (04:26 +0000)] 
DigestFetchState cbdata type moved to peer_digest.c

24 years agoMoved DigestFetchState cbdata definition to peer_digest.c and deleted
hno [Tue, 10 Apr 2001 03:55:50 +0000 (03:55 +0000)] 
Moved DigestFetchState cbdata definition to peer_digest.c and deleted
the old no longer used MEM_DIGEST_FETCH_STATE mempool.

24 years agochecklist->conn needs to be unlocked even if ident is not enabled. This
hno [Fri, 6 Apr 2001 07:49:58 +0000 (07:49 +0000)] 
checklist->conn needs to be unlocked even if ident is not enabled. This
because it is also used by the new proxy_auth model.

24 years agoDavid Eagles <David.EAGLES@ivolve.com.au> reports that cbdataInit()
wessels [Wed, 4 Apr 2001 06:42:36 +0000 (06:42 +0000)] 
David Eagles <David.EAGLES@ivolve.com.au> reports that cbdataInit()
is missing a call for CREATE_CBDATA(DigestFetchState).

24 years agoMake the location of the diskd program a configurable parameter.
adrian [Wed, 4 Apr 2001 02:22:10 +0000 (02:22 +0000)] 
Make the location of the diskd program a configurable parameter.

(YAY!).

This is configurable through the "diskd_program" entry in the squid
config file. It defaults to $(libexecdir). This means that
./configure --prefix="" --with-libexec-dir="" should now work properly.

24 years agoFurther to htcpHandleData problems, there was no case statment
wessels [Sun, 1 Apr 2001 22:52:31 +0000 (22:52 +0000)] 
Further to htcpHandleData problems, there was no case statment
for HTCP_CLR.

24 years agoin htcpHandleData() the check for htcp.opcode misses the case
wessels [Sun, 1 Apr 2001 22:33:37 +0000 (22:33 +0000)] 
in htcpHandleData() the check for htcp.opcode misses the case
when opcode equals HTCP_END.  This causes an assertion later
in the function.

24 years agoWhen squid is running in daemon mode, this patch causes the parent
wessels [Thu, 29 Mar 2001 06:24:18 +0000 (06:24 +0000)] 
When squid is running in daemon mode, this patch causes the parent
squid to exit if the child squid process was killed with SIGKILL.
Previously the parent would spawn a new child process, making it
difficult to stop squid.  We may want to add other signals to
the list as well.

24 years agoDW failed algebra
wessels [Wed, 28 Mar 2001 23:33:56 +0000 (23:33 +0000)] 
DW failed algebra

24 years agoSanity checking in storeDirGetBlkSize. Currently callers don't
wessels [Wed, 28 Mar 2001 21:45:40 +0000 (21:45 +0000)] 
Sanity checking in storeDirGetBlkSize.  Currently callers don't
check for errors or abnormal values.  Make sure fs.blksize is never
zero upon returning from this function.  Use 2048 as a default
block size.

24 years agoUpdate Eric's contact details
adrian [Tue, 27 Mar 2001 02:38:01 +0000 (02:38 +0000)] 
Update Eric's contact details

24 years agoFix a parameter which should be double instead of int.
adrian [Mon, 26 Mar 2001 01:08:25 +0000 (01:08 +0000)] 
Fix a parameter which should be double instead of int.
Its possible that CPU use could skyrocket if the compiler doesn't
do a friendly int -> double typecast for us, which it doesn't have
to..

Submitted by: Jürgen Sandner <juergen.sandner@baypol.bayern.de>

24 years agoHad the sign wrong on a recent patch to prevent negative entry->timestamp
wessels [Sat, 24 Mar 2001 03:19:11 +0000 (03:19 +0000)] 
Had the sign wrong on a recent patch to prevent negative entry->timestamp
values.

24 years agoBugfix to the previous patch. (kinkie)
hno [Thu, 22 Mar 2001 06:56:36 +0000 (06:56 +0000)] 
Bugfix to the previous patch. (kinkie)

24 years ago21 mar 2001 Robert Collins
hno [Thu, 22 Mar 2001 06:41:10 +0000 (06:41 +0000)] 
21 mar 2001 Robert Collins

* auth/ntlm/helpers/NTLMSSP/ntlm_auth.c: Send error details within
  the squid-helper protocol (Kinkie)
* auth/basic/auth_basic.c: Cleanup multiple requests when
  credentials recieved (Rob)
* acl.c: Use the connection credentials when testing for
  authentication status. (This fixes a recently introduced bug with
  persistent NTLM connections). (Rob)

24 years agocbdataFree rewritten to have a single thread of execution with no
hno [Tue, 20 Mar 2001 08:10:25 +0000 (08:10 +0000)] 
cbdataFree rewritten to have a single thread of execution with no
void data (only pointers).

24 years agoDavid Luyer's delay_pools patch from bugzilla #84.
wessels [Mon, 19 Mar 2001 13:18:42 +0000 (13:18 +0000)] 
David Luyer's delay_pools patch from bugzilla #84.

24 years agoFrom Robert Collins:
hno [Sat, 17 Mar 2001 20:31:16 +0000 (20:31 +0000)] 
From Robert Collins:

I've implemented pthread_cond* in cygwin (available in the 1.3.0
code - CVS HEAD branch), so the native win32 threaded code is no longer
needed. The in-cygwin implementation runs the existing aufs code nearly
unaltered. The attached patch brings in the same file  mode details
already implemented in ufs for cygwin (O_BINARY), not unlinking open
files... and #if tests the scheduler settings which aren't present on
cygwin yet.

24 years agofix delaypools/delay-pools typo and remove references to HEAP_REPLACEMENT
wessels [Thu, 15 Mar 2001 05:28:30 +0000 (05:28 +0000)] 
fix delaypools/delay-pools typo and remove references to HEAP_REPLACEMENT

24 years agoLooks like we forgot to return from dnsSubmit in the event of failure.
wessels [Wed, 14 Mar 2001 06:22:42 +0000 (06:22 +0000)] 
Looks like we forgot to return from dnsSubmit in the event of failure.

24 years agosd->log.open() and sd->log.close() are NULL for the "null" storage
wessels [Wed, 14 Mar 2001 02:11:25 +0000 (02:11 +0000)] 
sd->log.open() and sd->log.close() are NULL for the "null" storage
system.  The code was assuming that these functions are defined
for all storage types.  That caused coredumps upon reconfigure.  Now
only call those functions if non-NULL.

24 years agoran autoconf & autoheader
hno [Mon, 12 Mar 2001 05:54:17 +0000 (05:54 +0000)] 
ran autoconf & autoheader

24 years agoFrom Robert Collins:
hno [Mon, 12 Mar 2001 04:55:20 +0000 (04:55 +0000)] 
From Robert Collins:

The integer types patch introduced a problem with squid.h including
sys/types.h before defining FD_SETSIZE. This may cause problems on
IRIX, NETBSD, and definately does on cygwin ...

attached is a patch against HEAD to alter the location that the
squid_types.h is pulled in. I've done a make clean with this and it
was fine.

The changes to md5.c & h are because it depends on known int types.

24 years agoauth_rewrite update:
hno [Sat, 10 Mar 2001 07:55:35 +0000 (07:55 +0000)] 
auth_rewrite update:

Improved NTLMSSP debugging & reliability
Clearer acl.c logic, catches boundary cases.
"Fail open" configure option for NTLM to degrade gracefully
Should fix the compile issues on Solaris 8

Rob & kinkie.

24 years agocomments for hash_remove_link() are out of date.
wessels [Thu, 8 Mar 2001 00:57:37 +0000 (00:57 +0000)] 
comments for hash_remove_link() are out of date.

24 years agoHere are Gerben Wierda's files for installation and package
wessels [Tue, 6 Mar 2001 04:21:23 +0000 (04:21 +0000)] 
Here are Gerben Wierda's files for installation and package
maintenance on Nextstep.

24 years agoAdding these back for Gerben Wierda <Gerben_Wierda@rna.nl>
wessels [Tue, 6 Mar 2001 04:20:28 +0000 (04:20 +0000)] 
Adding these back for Gerben Wierda <Gerben_Wierda@rna.nl>

24 years agoundefined reference to `unlinkdUnlink' when --disable-unlinkd is used.
wessels [Sun, 4 Mar 2001 08:12:13 +0000 (08:12 +0000)] 
undefined reference to `unlinkdUnlink' when --disable-unlinkd is used.

24 years agoDW got the sign wrong on recent patch.
wessels [Sun, 4 Mar 2001 07:54:48 +0000 (07:54 +0000)] 
DW got the sign wrong on recent patch.

24 years agoForgot to assign the return value of logfileOpen() to useragentlog.
wessels [Sun, 4 Mar 2001 07:50:26 +0000 (07:50 +0000)] 
Forgot to assign the return value of logfileOpen() to useragentlog.

24 years agocbdata now have a per-type free function rather than per allocation.
hno [Sat, 3 Mar 2001 17:39:29 +0000 (17:39 +0000)] 
cbdata now have a per-type free function rather than per allocation.

Changes to the API:

To initialize a new CBDATA type with a free function:

   CBDATA_INIT_TYPE_FREECB(typename, freefunction);

To allocate a cbdata structure

   var = cbdataAlloc(typename);
   (was CBDATA_ALLOC(typename, freefunc) )

24 years agoAvoid NULL http->entry pointer in clientBuildReplyHeader().
wessels [Fri, 2 Mar 2001 06:02:31 +0000 (06:02 +0000)] 
Avoid NULL http->entry pointer in clientBuildReplyHeader().

24 years agodiskd clean log writing leaked one filedescriptor.
hno [Fri, 2 Mar 2001 05:28:22 +0000 (05:28 +0000)] 
diskd clean log writing leaked one filedescriptor.

24 years agoKilled one more unused variable.
hno [Fri, 2 Mar 2001 04:59:30 +0000 (04:59 +0000)] 
Killed one more unused variable.

24 years agoOops.. the transition to cache_dir options for Q1 and Q2 was not fully
hno [Fri, 2 Mar 2001 04:23:18 +0000 (04:23 +0000)] 
Oops.. the transition to cache_dir options for Q1 and Q2 was not fully
finished. There was many leftovers in the parsing code...

24 years agoFrom Costas Tavernarakis:
hno [Thu, 1 Mar 2001 10:27:54 +0000 (10:27 +0000)] 
From Costas Tavernarakis:

The following resolves squid 2.4 bug that on bugzilla is #79,
"NULL pointer access in storeDirSwapLog while writing icons".

ANSI C defines that integers in bitfields declared as 'int' are
either signed or unsigned, depending on the implementation.
gcc (unless -funsigned-bitfields is used) uses signed integers by
default. Sun cc uses unsinged.
Although not explicitly defined, squid expects the two bitfields
at lines 1324-1325 of structs.h to always be declared as signed,
which is not generally true.
Any portable program should always specify whether bitfield integer
types are signed or not.

24 years agoFrom Chemolli Francesco:
hno [Mon, 26 Feb 2001 04:49:07 +0000 (04:49 +0000)] 
From Chemolli Francesco:

Support for the types int16_t, int32_t, int64_t and their unsigned
counterparts u_int16_t, u_int32_t, u_int64_t. Legacy support for num32
is provided, but I recommend we deprecate num* types and (where struct
packing is useful) we move in time fully to the new types.

24 years agoFixed numerous compiler warnings found by IRIX MIPS compiler.
hno [Sat, 24 Feb 2001 03:59:50 +0000 (03:59 +0000)] 
Fixed numerous compiler warnings found by IRIX MIPS compiler.

24 years agoUndid indent messup (indent apparently cut this file into pieces)
hno [Sat, 24 Feb 2001 03:05:16 +0000 (03:05 +0000)] 
Undid indent messup (indent apparently cut this file into pieces)

24 years agoRemoving the "state == NULL" assertion in comm_write().
wessels [Sat, 24 Feb 2001 02:42:36 +0000 (02:42 +0000)] 
Removing the "state == NULL" assertion in comm_write().

Previously, this assertion was always true because we used
safe_free(state), but then state got MemPool-ified.  Although
we did NULL fd_table[fd].rwstate, we didn't NULL 'state'.

The assertion is unneeded because fd_table[fd].rwstate will
always be freed if it is set at the beginning of the function.

24 years agoediting bug: an extra #if from O_NONBLOCK change.
wessels [Fri, 23 Feb 2001 04:38:56 +0000 (04:38 +0000)] 
editing bug: an extra #if from O_NONBLOCK change.

24 years agoChanged the way that we pick between O_NDELAY and O_NONBLOCK for
wessels [Fri, 23 Feb 2001 04:05:56 +0000 (04:05 +0000)] 
Changed the way that we pick between O_NDELAY and O_NONBLOCK for
Solaris.  From way back we always used O_NDELAY rather than O_NONBLOCK
for SunOS and Solaris.  The actual reasons why we assumed O_NONBLOCK is
broken are forgotten.

Recent evidence indicates that O_NONBLOCK on Solaris (v8) works
correctly and is actually required for detecting EOF on kernel
pipes (e.g. with diskd).

This change will make solaris use O_NONBLOCK if its defined.

24 years agoAvoid making 'served_date' negative in storeTimestampsSet().
wessels [Fri, 23 Feb 2001 03:53:15 +0000 (03:53 +0000)] 
Avoid making 'served_date' negative in storeTimestampsSet().

24 years agocachemgr fix for http_header_access and header_replacement
hno [Wed, 21 Feb 2001 07:02:34 +0000 (07:02 +0000)] 
cachemgr fix for http_header_access and header_replacement
by Robert Collins

Specifically neither produce output that can be put into a squid.conf
file...

24 years agopipeline_prefetch directive to disable/enable the prefetch of pipelined
hno [Wed, 21 Feb 2001 05:49:23 +0000 (05:49 +0000)] 
pipeline_prefetch directive to disable/enable the prefetch of pipelined
requests.

24 years agoDon't try to remove memory-only objects from the disk during cache_dir rebuild
hno [Tue, 20 Feb 2001 06:10:06 +0000 (06:10 +0000)] 
Don't try to remove memory-only objects from the disk during cache_dir rebuild

24 years agoRemoved the "default:" case from most ACL-type switch statements to get
hno [Sun, 18 Feb 2001 18:16:51 +0000 (18:16 +0000)] 
Removed the "default:" case from most ACL-type switch statements to get
warnings from GCC when we forget one of the types.

This found problems with

   ACL_REQ_MIME_TYPE
   ACL_REP_MIME_TYPE
   ACL_NETDB_SRC_RTT (not actually implemented it seems)

24 years agoDon't use size_t in the wrong context. size_t is unsigned, and not good
hno [Sat, 17 Feb 2001 14:30:44 +0000 (14:30 +0000)] 
Don't use size_t in the wrong context. size_t is unsigned, and not good
for storing syscall return values which always are signed to cover errors..

24 years agoRemoved some old "pump" leftovers
hno [Sat, 17 Feb 2001 14:28:50 +0000 (14:28 +0000)] 
Removed some old "pump" leftovers

24 years agoPrompt people to send email to squid-bugs@squid-cache.org rather than
adrian [Thu, 15 Feb 2001 18:11:54 +0000 (18:11 +0000)] 
Prompt people to send email to squid-bugs@squid-cache.org rather than
squid-bugs@ircache.net.

24 years agoDetect 0.0.0.0 DNS server specifications, warn the user and use
hno [Thu, 15 Feb 2001 03:58:10 +0000 (03:58 +0000)] 
Detect 0.0.0.0 DNS server specifications, warn the user and use
127.0.0.1 instead.

24 years agoFrom Robert Collins:
hno [Wed, 14 Feb 2001 04:45:48 +0000 (04:45 +0000)] 
From Robert Collins:

The bug is really more of an unexpected feature: when testing for
content-type on reply mime headers, messages with no entitie will
never match a mime type, so restrictive access configurations will
not let them though. The bad thing is that 304 responses (along
with 1xx & 204) will never have a content-type header. Therefore
they never match and IMS responses to the client always fail. And
at the moment there is no acl for the status line, so we cannot
let the user decide...

24 years agoCleanup of how "httpd_accel_port virtual" is handled. Now httpd_accel_port
hno [Mon, 12 Feb 2001 03:06:15 +0000 (03:06 +0000)] 
Cleanup of how "httpd_accel_port virtual" is handled. Now httpd_accel_port
is always (for accelerated request) used if it is set, and if virtual then
the port number is retreived from
a) Host header (when httpd_accel_uses_host_header is on)
b) Transparent redirection/NAT lookup
c) Local endpoint of the socket

Previosly is was a mix where some always overrode httpd_accel_port, and
in some cases virtual was even ignored causing a port number of 0..

24 years agoRemoved some old stale "todo" comments about EWOULDBLOCK which no longer
hno [Sun, 11 Feb 2001 21:10:34 +0000 (21:10 +0000)] 
Removed some old stale "todo" comments about EWOULDBLOCK which no longer
applies (async-io never returns EWOULDBLOCK on overload in the current
implementation)

24 years agoSerbian error pages by Zoran Verovski <Zoran@HEMOFARM.CO.YU>
hno [Sun, 11 Feb 2001 09:54:22 +0000 (09:54 +0000)] 
Serbian error pages by Zoran Verovski <Zoran@HEMOFARM.CO.YU>

24 years agoThe "maxobjsize" parameter moved to an optional cache_dir option.
hno [Sat, 10 Feb 2001 23:40:40 +0000 (23:40 +0000)] 
The "maxobjsize" parameter moved to an optional cache_dir option.

This was done by implementing a generic option parser for cache_dir
where each FS-type can include their own options if needed.

squid.conf syntax changes:

maxobjsize argument now an option
the Q1 and Q2 arguments to diskd now options

Generic cache_dir syntax

  cache_dir type path [fs-parameters] [optins]

Common options available to all stores

   max-size=n (the former maxobjsize argument)
   read-only

diskd options
   Q1=n, Q2=n (the former Q1 and Q2 arguments)

Implementation details:

Each option is defined by a struct cache_dir_option with the members
  name   option name
  parse  function that parses the option argument

The parse functions are called with the arguments
  parse_function(SwapDir *sd, const char *optionname,
  const char *optionvalue, int reconfiguring)

The FS cache_dir parser (both initial and reconfigure) must call
parse_cachedir_options after parsing all it's FS-specific arguments.

  parse_cachedir_options(SwapDir *sd, struct cache_dir_option *options,
  int reconfiguring)

options is an array of FS-specific options, terminated with a entry where
name is NULL.

The list of common options shared by all stores is defined in cache_cf.c

24 years agoSpelling correction in a command which could get triggered in parallell builds
hno [Sat, 10 Feb 2001 21:39:33 +0000 (21:39 +0000)] 
Spelling correction in a command which could get triggered in parallell builds

24 years agoFrom Robert Collins:
hno [Sat, 10 Feb 2001 02:35:10 +0000 (02:35 +0000)] 
From Robert Collins:

implements a new acl type rep_mime_type, complementary to
req_mime_type, and a new access list http_reply_access which allows acls
to be applied to replies. The prime target is content-type filtering.
One such example would be to only allow a particular set of users the
ability to download tgz. (ie by combining proxy_auth and rep_mime_type).

I created a new acl type rather than moving the request vs reply header
field decision to the calling routine as that seems to be the approach
all over the place.

24 years agoCleaned up the namespace (local functions made static)
hno [Thu, 8 Feb 2001 02:10:12 +0000 (02:10 +0000)] 
Cleaned up the namespace (local functions made static)

24 years agoOops... was a bit tired there.
hno [Thu, 8 Feb 2001 02:09:25 +0000 (02:09 +0000)] 
Oops... was a bit tired there.

24 years agoCleaned up the namespace (local functions made static)
hno [Thu, 8 Feb 2001 02:04:04 +0000 (02:04 +0000)] 
Cleaned up the namespace (local functions made static)

24 years agoFix for a forgotten cbdata change in digest generation
hno [Thu, 8 Feb 2001 02:00:10 +0000 (02:00 +0000)] 
Fix for a forgotten cbdata change in digest generation

24 years agoCleaned up the namespace (local functions made static)
hno [Thu, 8 Feb 2001 01:56:50 +0000 (01:56 +0000)] 
Cleaned up the namespace (local functions made static)

24 years agoCleaned up the "null" FS by removing all unused "junk", and merging it into
hno [Thu, 8 Feb 2001 00:42:21 +0000 (00:42 +0000)] 
Cleaned up the "null" FS by removing all unused "junk", and merging it into
one (small) file.

24 years agoUppered the metadata swapin element length from 1<<10 to 1<<16 bytes. This
hno [Fri, 2 Feb 2001 02:48:25 +0000 (02:48 +0000)] 
Uppered the metadata swapin element length from 1<<10 to 1<<16 bytes. This
was causing bogous swapin warnings on long URL's (>1024 characters).

24 years agoRemoved unused "retries" variable
hno [Fri, 2 Feb 2001 02:00:59 +0000 (02:00 +0000)] 
Removed unused "retries" variable

24 years agosleep() needs unistd.h
hno [Fri, 2 Feb 2001 01:59:37 +0000 (01:59 +0000)] 
sleep() needs unistd.h

24 years agoran autoconf
hno [Fri, 2 Feb 2001 01:51:52 +0000 (01:51 +0000)] 
ran autoconf

24 years agoUse strcasecmp, not stricmp
hno [Fri, 2 Feb 2001 01:49:06 +0000 (01:49 +0000)] 
Use strcasecmp, not stricmp

24 years agoRemoved duplicate entry for ACL_REQ_MIME_TYPE
hno [Fri, 2 Feb 2001 01:47:46 +0000 (01:47 +0000)] 
Removed duplicate entry for ACL_REQ_MIME_TYPE

24 years agoMajor update from auth_rewrite
hno [Thu, 1 Feb 2001 05:16:35 +0000 (05:16 +0000)] 
Major update from auth_rewrite

Robert Collins:

Digest (RFC2617) proxy authentication implementation

Chemolli Francesco:

Several bugs in NTLM authentication when dealing with untrusted
domains, wrong passwords, helper arguments and more have been fixed

24 years agoacl req_mime_type destroy patch from Robert.
hno [Thu, 1 Feb 2001 04:48:24 +0000 (04:48 +0000)] 
acl req_mime_type destroy patch from Robert.

24 years agooops.. forgot /g in the substitution.
hno [Thu, 1 Feb 2001 04:46:04 +0000 (04:46 +0000)] 
oops.. forgot /g in the substitution.

24 years agos/unlock_func/free_func/ in cbdata.c to more closely match the meaning...
hno [Thu, 1 Feb 2001 04:44:42 +0000 (04:44 +0000)] 
s/unlock_func/free_func/ in cbdata.c to more closely match the meaning...

24 years agoOops.. the coredump_dir patch was incomplete, with quite obvious errors
hno [Wed, 31 Jan 2001 07:25:59 +0000 (07:25 +0000)] 
Oops.. the coredump_dir patch was incomplete, with quite obvious errors
(did not even compile). Shame on me.

24 years agomaximum_icp_query_timeout comment fix. Found by Toby Dickenson.
hno [Wed, 31 Jan 2001 04:31:05 +0000 (04:31 +0000)] 
maximum_icp_query_timeout comment fix. Found by Toby Dickenson.

24 years agoCleanup of coredump_dir. No longer automatically uses the first cache_dir
hno [Tue, 30 Jan 2001 17:06:48 +0000 (17:06 +0000)] 
Cleanup of coredump_dir. No longer automatically uses the first cache_dir
location, and can be commented out to use the current directory.

24 years agoMinor cleanup of parse_eol to get rid of leading whitespace
hno [Tue, 30 Jan 2001 16:46:46 +0000 (16:46 +0000)] 
Minor cleanup of parse_eol to get rid of leading whitespace

24 years agoFrom Robert Collins:
hno [Sun, 28 Jan 2001 23:23:36 +0000 (23:23 +0000)] 
From Robert Collins:
The req_mime_type acl was  parsed as a word list and matched as a regex :-/

24 years agoAllow for URL escapes in the login:password string. This to enable the use
hno [Sun, 28 Jan 2001 07:45:45 +0000 (07:45 +0000)] 
Allow for URL escapes in the login:password string. This to enable the use
of users or passwords with spaces in them (%20)

24 years agoOops.. an extra d slipped in in the commit.
hno [Fri, 26 Jan 2001 06:01:56 +0000 (06:01 +0000)] 
Oops.. an extra d slipped in in the commit.

24 years agoFrom Kevin Littlejohn:
hno [Fri, 26 Jan 2001 05:54:52 +0000 (05:54 +0000)] 
From Kevin Littlejohn:

makes cache_cf.c not try and figure out nobody and nogroup if it's not
going to use them - ie. if squid is being run as a user instead of as root.
It came up because I don't have a nogroup on my machine.

Additions by Henrik Nordstrom:

cache_effective_group now defaults to the GID of cache_effective_user, or
the current user if not started as root. If can always be specified to some
other value if required, but are then subject to the permissions set by the
os (i.e. only root can set the group to a group where he/she is not a member)

24 years agocacheCurrentResFileDescrCnt was reporting "Number_FD" rather than
wessels [Thu, 25 Jan 2001 05:38:25 +0000 (05:38 +0000)] 
cacheCurrentResFileDescrCnt was reporting "Number_FD" rather than
RESERVED_FD.  Note the MIB doesn't have a variable for the number
of filedescriptors currently open.

24 years agoFix storeDirDiskdCheckObj. The loadav calculation was backwards.
wessels [Wed, 24 Jan 2001 00:27:21 +0000 (00:27 +0000)] 
Fix storeDirDiskdCheckObj.  The loadav calculation was backwards.
Also removed some old, unused code.

24 years agoKilled some extra \"
hno [Tue, 23 Jan 2001 08:07:19 +0000 (08:07 +0000)] 
Killed some extra \"

24 years agoOops.. don't do the translation in non-HTML files.
hno [Tue, 23 Jan 2001 08:06:06 +0000 (08:06 +0000)] 
Oops.. don't do the translation in non-HTML files.

24 years agoTranslated from 8-bit to use HTML entity names for the national symbols
hno [Tue, 23 Jan 2001 08:05:16 +0000 (08:05 +0000)] 
Translated from 8-bit to use HTML entity names for the national symbols

24 years agodrees@spoke.nols.com found some typos.
wessels [Fri, 19 Jan 2001 00:57:04 +0000 (00:57 +0000)] 
drees@spoke.nols.com found some typos.

24 years agoUse "mkdir -p" to create directories
hno [Wed, 17 Jan 2001 04:11:23 +0000 (04:11 +0000)] 
Use "mkdir -p" to create directories

24 years agoMake sure they sign bit of 'ikey' is never set so the diskd IPC keys
wessels [Tue, 16 Jan 2001 05:37:10 +0000 (05:37 +0000)] 
Make sure they sign bit of 'ikey' is never set so the diskd IPC keys
remain positive.

24 years agoChanged
wessels [Tue, 16 Jan 2001 01:17:00 +0000 (01:17 +0000)] 
Changed
sed -e'foo'
to
sed -e 'foo'
because the first appears to not work on some solaris boxes