platform/upstream/curl.git
20 years agoDon't check for HAVE_ defines in this header file, it is meant to be public
Daniel Stenberg [Wed, 25 Feb 2004 07:22:00 +0000 (07:22 +0000)]
Don't check for HAVE_ defines in this header file, it is meant to be public
and we can't depend on configure-defines in it. This logic is borrowed from
the public curl headers.

20 years agofix
Daniel Stenberg [Wed, 25 Feb 2004 07:17:31 +0000 (07:17 +0000)]
fix

20 years agoDan Fandrich fixed a minor flaw in Dominick's fix!
Daniel Stenberg [Wed, 25 Feb 2004 06:37:42 +0000 (06:37 +0000)]
Dan Fandrich fixed a minor flaw in Dominick's fix!

20 years agocheck for a few basic header files
Daniel Stenberg [Mon, 23 Feb 2004 16:20:42 +0000 (16:20 +0000)]
check for a few basic header files

20 years agoinclude sys/select.h
Daniel Stenberg [Mon, 23 Feb 2004 16:20:31 +0000 (16:20 +0000)]
include sys/select.h

20 years agofixed some warnings in the (both new and old) base64 usage
Daniel Stenberg [Mon, 23 Feb 2004 16:09:39 +0000 (16:09 +0000)]
fixed some warnings in the (both new and old) base64 usage

20 years agofeatures no one seem to care much about are now moved over to the 7.11.2
Daniel Stenberg [Mon, 23 Feb 2004 16:04:16 +0000 (16:04 +0000)]
features no one seem to care much about are now moved over to the 7.11.2
release

20 years agomake newer MSCV7 compilers use _strtoi64() as a strtoll() replacement
Daniel Stenberg [Mon, 23 Feb 2004 14:24:49 +0000 (14:24 +0000)]
make newer MSCV7 compilers use _strtoi64() as a strtoll() replacement

20 years agolimit rate and windows timeouts
Daniel Stenberg [Mon, 23 Feb 2004 13:48:27 +0000 (13:48 +0000)]
limit rate and windows timeouts

20 years agoares_free_errmem is gone
Daniel Stenberg [Mon, 23 Feb 2004 13:35:28 +0000 (13:35 +0000)]
ares_free_errmem is gone

20 years agounused and now removed
Daniel Stenberg [Mon, 23 Feb 2004 13:33:18 +0000 (13:33 +0000)]
unused and now removed

20 years agoRemoved getpass from here. The fact this still was present here indicates
Daniel Stenberg [Mon, 23 Feb 2004 12:01:10 +0000 (12:01 +0000)]
Removed getpass from here. The fact this still was present here indicates
that this file is out of date!

20 years agocurl_off_t is 8 bytes big on windows
Daniel Stenberg [Mon, 23 Feb 2004 11:59:28 +0000 (11:59 +0000)]
curl_off_t is 8 bytes big on windows

20 years agoswitch the arguments according to the c source
Daniel Stenberg [Mon, 23 Feb 2004 11:44:03 +0000 (11:44 +0000)]
switch the arguments according to the c source

20 years agooops, the decode() function got its arguments reversed in my cleanup
Daniel Stenberg [Mon, 23 Feb 2004 11:39:21 +0000 (11:39 +0000)]
oops, the decode() function got its arguments reversed in my cleanup
operation!

20 years agocorrect the input data structs
Daniel Stenberg [Mon, 23 Feb 2004 09:01:08 +0000 (09:01 +0000)]
correct the input data structs

20 years agouse size_t to keep strlen() results
Daniel Stenberg [Mon, 23 Feb 2004 08:38:49 +0000 (08:38 +0000)]
use size_t to keep strlen() results

20 years agosimplied how create_hostcache_id() is used, and also its function somewhat
Daniel Stenberg [Mon, 23 Feb 2004 08:35:57 +0000 (08:35 +0000)]
simplied how create_hostcache_id() is used, and also its function somewhat
cleared up some ssize_t/size_t mixups

20 years agoadjusted to the modified base64 protos
Daniel Stenberg [Mon, 23 Feb 2004 08:24:53 +0000 (08:24 +0000)]
adjusted to the modified base64 protos

20 years agoMore size_t cleanups in the base64 functions.
Daniel Stenberg [Mon, 23 Feb 2004 08:22:43 +0000 (08:22 +0000)]
More size_t cleanups in the base64 functions.

20 years agoThe base64 encode function now takes a size_t for size, not an int as
Daniel Stenberg [Mon, 23 Feb 2004 08:07:55 +0000 (08:07 +0000)]
The base64 encode function now takes a size_t for size, not an int as
previously.

20 years agomore int vs long/size_t fixes after icc compiler warnings
Daniel Stenberg [Mon, 23 Feb 2004 08:04:31 +0000 (08:04 +0000)]
more int vs long/size_t fixes after icc compiler warnings

20 years agomore int/long fixes after icc "remarks"
Daniel Stenberg [Mon, 23 Feb 2004 08:00:56 +0000 (08:00 +0000)]
more int/long fixes after icc "remarks"

20 years agonext_id is an unsigned short, typecast the assign to prevent picky compilers
Daniel Stenberg [Mon, 23 Feb 2004 07:57:44 +0000 (07:57 +0000)]
next_id is an unsigned short, typecast the assign to prevent picky compilers
to warn

20 years agotypecase the bitfiddling results since we get a long and we store an int,
Daniel Stenberg [Mon, 23 Feb 2004 07:55:58 +0000 (07:55 +0000)]
typecase the bitfiddling results since we get a long and we store an int,
they may not be of the size size

20 years agomake ares_expand_name() take a long * instead of an int *, since we do
Daniel Stenberg [Mon, 23 Feb 2004 07:52:20 +0000 (07:52 +0000)]
make ares_expand_name() take a long * instead of an int *, since we do
pointer arithmetic (ptr1 - ptr2) and to do that properly on 64bit we need
long

20 years agostrlen() returns a size_t, which might be larger than int on some platforms
Daniel Stenberg [Mon, 23 Feb 2004 07:46:59 +0000 (07:46 +0000)]
strlen() returns a size_t, which might be larger than int on some platforms

20 years agoargh, use single-quotes instead of double ones to make $Revision appear
Daniel Stenberg [Mon, 23 Feb 2004 07:32:45 +0000 (07:32 +0000)]
argh, use single-quotes instead of double ones to make $Revision appear

20 years agouse the proper source header, and set the version string to this file's
Daniel Stenberg [Mon, 23 Feb 2004 07:17:13 +0000 (07:17 +0000)]
use the proper source header, and set the version string to this file's
revision number

20 years agocd back to the root path before removing the build dir, since some systems
Daniel Stenberg [Mon, 23 Feb 2004 07:08:59 +0000 (07:08 +0000)]
cd back to the root path before removing the build dir, since some systems
refuse to remove the dir otherwise!

20 years agoDominick Meglio's fix for supporting multiple names in the Nameserver key
Daniel Stenberg [Sun, 22 Feb 2004 22:42:20 +0000 (22:42 +0000)]
Dominick Meglio's fix for supporting multiple names in the Nameserver key
on Windows.

20 years agothis code uses no long long, so we can have warnings about them
Daniel Stenberg [Sun, 22 Feb 2004 22:36:36 +0000 (22:36 +0000)]
this code uses no long long, so we can have warnings about them

20 years agothe missing part of Gisle Vanem's connect-timeout fix for win32
Daniel Stenberg [Sun, 22 Feb 2004 22:31:24 +0000 (22:31 +0000)]
the missing part of Gisle Vanem's connect-timeout fix for win32

20 years ago-Wno-format-nonliteral does not exist in my gcc 2.96, only use that with
Daniel Stenberg [Sat, 21 Feb 2004 16:56:01 +0000 (16:56 +0000)]
-Wno-format-nonliteral does not exist in my gcc 2.96, only use that with
newer versions

20 years agomention in --limit-rate that --speed-limit might ruin the limiting slightly.
Daniel Stenberg [Sat, 21 Feb 2004 16:18:57 +0000 (16:18 +0000)]
mention in --limit-rate that --speed-limit might ruin the limiting slightly.

20 years agoDavid Byron's fix to allow the speed-limit logic work even if you set
Daniel Stenberg [Sat, 21 Feb 2004 15:08:21 +0000 (15:08 +0000)]
David Byron's fix to allow the speed-limit logic work even if you set
limit-rate. It does work on the expense of the rate limiter.

20 years agoadded some extra typecasts to prevent compiler warnings when converting
Daniel Stenberg [Sat, 21 Feb 2004 15:05:46 +0000 (15:05 +0000)]
added some extra typecasts to prevent compiler warnings when converting
int to various types

20 years agoinlcude krb4.h to get the proto for Curl_krb_kauth() to satisfy picky
Daniel Stenberg [Sat, 21 Feb 2004 14:57:29 +0000 (14:57 +0000)]
inlcude krb4.h to get the proto for Curl_krb_kauth() to satisfy picky
compilers

20 years agoGisle Vanem brings name resolving timeout possibilities to windows people.
Daniel Stenberg [Fri, 20 Feb 2004 16:41:50 +0000 (16:41 +0000)]
Gisle Vanem brings name resolving timeout possibilities to windows people.
This works by magicly starting up a new thread that can be killed when the
timeout is reached. testtesttest!

20 years agothat icc attempt in configure
Daniel Stenberg [Fri, 20 Feb 2004 16:29:57 +0000 (16:29 +0000)]
that icc attempt in configure

20 years agoSome compilers warn on completely empty source files, we provide a blank
Daniel Stenberg [Fri, 20 Feb 2004 16:22:47 +0000 (16:22 +0000)]
Some compilers warn on completely empty source files, we provide a blank
one to prevent that.

20 years agoConvert functions to ANSI-style declaration to prevent compiler warnings
Daniel Stenberg [Fri, 20 Feb 2004 16:18:26 +0000 (16:18 +0000)]
Convert functions to ANSI-style declaration to prevent compiler warnings

20 years agoWhen the built-in manual is disabled, we generate a function doing nothing
Daniel Stenberg [Fri, 20 Feb 2004 15:39:37 +0000 (15:39 +0000)]
When the built-in manual is disabled, we generate a function doing nothing
just to avoid making a totally empty file. Just to avoid compiler warnings.

20 years agoNo longer support Z as a flag to print size_t, it isn't used by libcurl
Daniel Stenberg [Fri, 20 Feb 2004 15:16:31 +0000 (15:16 +0000)]
No longer support Z as a flag to print size_t, it isn't used by libcurl
and I doubt anyone else uses it.

Better preprocessor magic for the O flag (for curl_off_t printing) to prevent
compiler warnings.

20 years agocool.haxx.se, no longer sourceforge
Daniel Stenberg [Fri, 20 Feb 2004 13:09:35 +0000 (13:09 +0000)]
cool.haxx.se, no longer sourceforge

20 years agoOnly use -Wstrict-prototypes with gcc 3.3 or later.
Daniel Stenberg [Fri, 20 Feb 2004 10:11:50 +0000 (10:11 +0000)]
Only use -Wstrict-prototypes with gcc 3.3 or later.
It is working with earlier versions, but when I use it with 3.0.3, I can't
get it to ignore errors in "system headers" with -isystem so we get
excessive amounts of warnings on SSL headers which is very annoying.

20 years agoinclude sys/select.h to prevent picky compiler warnings when using select()
Daniel Stenberg [Fri, 20 Feb 2004 08:51:43 +0000 (08:51 +0000)]
include sys/select.h to prevent picky compiler warnings when using select()
without proto

20 years agowe call the macro CURLMAX() isntead of MAX(), just because it turned up
Daniel Stenberg [Fri, 20 Feb 2004 08:47:23 +0000 (08:47 +0000)]
we call the macro CURLMAX() isntead of MAX(), just because it turned up
it collides with another MAX define on some platforms (like netbsd 1.6.1)

20 years agooops, missed a define when I changed from Curl_ to curlx_
Daniel Stenberg [Fri, 20 Feb 2004 07:22:25 +0000 (07:22 +0000)]
oops, missed a define when I changed from Curl_ to curlx_

20 years agofix protos to prevent warnings
Daniel Stenberg [Fri, 20 Feb 2004 07:19:18 +0000 (07:19 +0000)]
fix protos to prevent warnings

20 years agoadded getpart.h for the spitout() proto
Daniel Stenberg [Fri, 20 Feb 2004 07:19:01 +0000 (07:19 +0000)]
added getpart.h for the spitout() proto

20 years ago%HTTPPORT supported in subVariables
Daniel Stenberg [Fri, 20 Feb 2004 07:14:58 +0000 (07:14 +0000)]
%HTTPPORT supported in subVariables

20 years agomention what ports the test suite uses
Daniel Stenberg [Fri, 20 Feb 2004 07:05:10 +0000 (07:05 +0000)]
mention what ports the test suite uses

20 years agoWhen trying to see if there's a friendly http server on "our" port, we
Daniel Stenberg [Fri, 20 Feb 2004 06:59:17 +0000 (06:59 +0000)]
When trying to see if there's a friendly http server on "our" port, we
only accept return code 7 to indicate that there's no server present.

20 years ago-Wcast-align is a bit too annoying
Daniel Stenberg [Thu, 19 Feb 2004 21:34:52 +0000 (21:34 +0000)]
-Wcast-align is a bit too annoying

20 years agoAnything that looks like gcc 5.0 or more is no longer treated as gcc. I hope
Daniel Stenberg [Thu, 19 Feb 2004 21:32:19 +0000 (21:32 +0000)]
Anything that looks like gcc 5.0 or more is no longer treated as gcc. I hope
this will make us exclude icc 8.0 etc.

20 years agorefer to FILEFORMAT
Daniel Stenberg [Thu, 19 Feb 2004 21:21:12 +0000 (21:21 +0000)]
refer to FILEFORMAT

20 years agoWhen --enable-debug is used, for every -I provided to $CPPFLAGS we add
Daniel Stenberg [Thu, 19 Feb 2004 19:25:11 +0000 (19:25 +0000)]
When --enable-debug is used, for every -I provided to $CPPFLAGS we add
a corresponding -isystem, if using gcc, to inhibit warnings on those
headers.

20 years agoTor Arntsen's tiny fix!
Daniel Stenberg [Thu, 19 Feb 2004 16:24:22 +0000 (16:24 +0000)]
Tor Arntsen's tiny fix!

20 years agoif not yacc or bison is found, check if we *really* need it, and if we do
Daniel Stenberg [Thu, 19 Feb 2004 15:58:58 +0000 (15:58 +0000)]
if not yacc or bison is found, check if we *really* need it, and if we do
we bail out!

20 years agofixed the no-ssl version to return int as well
Daniel Stenberg [Thu, 19 Feb 2004 15:39:06 +0000 (15:39 +0000)]
fixed the no-ssl version to return int as well

20 years agoprovide protos to the functions to prevent warnings
Daniel Stenberg [Thu, 19 Feb 2004 13:03:04 +0000 (13:03 +0000)]
provide protos to the functions to prevent warnings

20 years agotypecast the type to an int on return
Daniel Stenberg [Thu, 19 Feb 2004 13:00:33 +0000 (13:00 +0000)]
typecast the type to an int on return

20 years agoreturn int from test()
Daniel Stenberg [Thu, 19 Feb 2004 13:00:05 +0000 (13:00 +0000)]
return int from test()

20 years agoprovide a test() proto
Daniel Stenberg [Thu, 19 Feb 2004 12:59:57 +0000 (12:59 +0000)]
provide a test() proto

20 years agouse a die and a log function to die and log texts
Daniel Stenberg [Thu, 19 Feb 2004 12:56:45 +0000 (12:56 +0000)]
use a die and a log function to die and log texts
better detect test suite failures
remove the buildlog at exit
make a random buildlog file name, now in the same dir where the build dir
is created
checks if the ares build succeeded

20 years agoremove the build.log too at exit, and also use the proper $pwd prefix
Daniel Stenberg [Thu, 19 Feb 2004 12:10:07 +0000 (12:10 +0000)]
remove the build.log too at exit, and also use the proper $pwd prefix
to find the files/dirs to remove so that it still works if we "die" after
having done a 'cd'

20 years agovery minor phrase edit
Daniel Stenberg [Thu, 19 Feb 2004 10:21:54 +0000 (10:21 +0000)]
very minor phrase edit

20 years agofixes from the recent days
Daniel Stenberg [Thu, 19 Feb 2004 10:21:43 +0000 (10:21 +0000)]
fixes from the recent days

20 years agocheck for standard headers when --enable-debug is used
Daniel Stenberg [Thu, 19 Feb 2004 09:33:59 +0000 (09:33 +0000)]
check for standard headers when --enable-debug is used

20 years agoDoug Porter's patch that changes the order of preferences on how to find the
Daniel Stenberg [Thu, 19 Feb 2004 09:22:00 +0000 (09:22 +0000)]
Doug Porter's patch that changes the order of preferences on how to find the
default netrc file. We now read and uses HOME _before_ we use getpwuid() to
better allow users to move around HOME to use different .netrc files without
having to rely on even blacker magic.

20 years agoIf --enable-debug is used and gcc, we figure out which version and then we
Daniel Stenberg [Thu, 19 Feb 2004 09:01:13 +0000 (09:01 +0000)]
If --enable-debug is used and gcc, we figure out which version and then we
use as aggressive warning options as possible for the used compiler version.

20 years agoitem 19 is considered fixed until we get to hear differently
Daniel Stenberg [Thu, 19 Feb 2004 08:19:21 +0000 (08:19 +0000)]
item 19 is considered fixed until we get to hear differently
item 25 is now dealt with using the curlx_ prefix, mentioned in detail here:
http://curl.haxx.se/mail/lib-2004-02/0215.html

20 years agoUse the strtoofft.h header file from the lib directory, as we are now
Daniel Stenberg [Thu, 19 Feb 2004 08:13:20 +0000 (08:13 +0000)]
Use the strtoofft.h header file from the lib directory, as we are now
officially using library-code when building the app (at least for the platforms
that don't have a strtoll() on their own).

20 years agoRemade to use curlx_-prefix. This means this function can be compiled and
Daniel Stenberg [Thu, 19 Feb 2004 08:12:13 +0000 (08:12 +0000)]
Remade to use curlx_-prefix. This means this function can be compiled and
linked separately by the application. This function is not provided by the
libcurl API. It can only be accessed by apps if they compile and use this
particular source code.

20 years agoAIX and Tru64 have what Tor calls "horribly broken 'which' programs" so we
Daniel Stenberg [Wed, 18 Feb 2004 16:16:13 +0000 (16:16 +0000)]
AIX and Tru64 have what Tor calls "horribly broken 'which' programs" so we
now scan the PATH ourself to find the path to (g)libtool

20 years agoremoved some "jhrg" from comments
Daniel Stenberg [Wed, 18 Feb 2004 15:28:56 +0000 (15:28 +0000)]
removed some "jhrg" from comments

20 years agoI removed the socklen_t requirement from memdebug.h, so we don't need to
Daniel Stenberg [Wed, 18 Feb 2004 12:26:27 +0000 (12:26 +0000)]
I removed the socklen_t requirement from memdebug.h, so we don't need to
figure it out here anymore to build debug builds.

20 years agoMade curl_accept() take a 'void *' instead of 'socklen_t *' in the 3rd
Daniel Stenberg [Wed, 18 Feb 2004 12:22:56 +0000 (12:22 +0000)]
Made curl_accept() take a 'void *' instead of 'socklen_t *' in the 3rd
argument to also not force the casual includer to know about the socklen_t
type.

20 years agoModified curl_accept() to take a 'void *' in the 2nd argument instead of
Daniel Stenberg [Wed, 18 Feb 2004 12:18:33 +0000 (12:18 +0000)]
Modified curl_accept() to take a 'void *' in the 2nd argument instead of
sockaddr *. This has the added benefit that source files that include
memdebug.h doesn't have to know about "sockaddr".

20 years agoNo longer uses the 'ret' variable in the plain ipv4-version of
Daniel Stenberg [Wed, 18 Feb 2004 10:05:17 +0000 (10:05 +0000)]
No longer uses the 'ret' variable in the plain ipv4-version of
my_getaddrinfo() (caused a warning by the IRIX MIPSPro compiler). Also
clarified the situation for the 3-arg version of gethostbyname_r() with a huge
comment.

20 years agoThe --enable-debug option really requires this to be built as part of curl.
Daniel Stenberg [Wed, 18 Feb 2004 09:07:50 +0000 (09:07 +0000)]
The --enable-debug option really requires this to be built as part of curl.
When using it, we now set the include path to better find the devel curl
headers, and we check for the socklen_t type since the curl memdebug stuff
needs it.

20 years agosimplified and better commented config.h include logic
Daniel Stenberg [Wed, 18 Feb 2004 08:35:51 +0000 (08:35 +0000)]
simplified and better commented config.h include logic

20 years agoMake sure dns cache timeout -1 really means forever, as it is documented to
Daniel Stenberg [Wed, 18 Feb 2004 07:56:18 +0000 (07:56 +0000)]
Make sure dns cache timeout -1 really means forever, as it is documented to
be. Simply skip the pruning.

20 years agofix the help text for --manual if built without manual
Daniel Stenberg [Tue, 17 Feb 2004 13:46:00 +0000 (13:46 +0000)]
fix the help text for --manual if built without manual

20 years agoignore more
Daniel Stenberg [Tue, 17 Feb 2004 07:57:31 +0000 (07:57 +0000)]
ignore more

20 years agomemdebug build, 'make' no longer builds the demo tools
Daniel Stenberg [Tue, 17 Feb 2004 07:41:10 +0000 (07:41 +0000)]
memdebug build, 'make' no longer builds the demo tools

20 years ago'make all' also builds the demos
Daniel Stenberg [Tue, 17 Feb 2004 07:40:46 +0000 (07:40 +0000)]
'make all' also builds the demos

20 years agoinclude ares_private.h to make sure we get the memdebug stuff included
Daniel Stenberg [Tue, 17 Feb 2004 07:40:31 +0000 (07:40 +0000)]
include ares_private.h to make sure we get the memdebug stuff included

20 years agoIf CURLDEBUG is set we use the libcurl internal memdebug system to track
Daniel Stenberg [Tue, 17 Feb 2004 07:40:02 +0000 (07:40 +0000)]
If CURLDEBUG is set we use the libcurl internal memdebug system to track
memory leaks etc.

20 years agoonly build adig and ahost if 'make demos' is used
Daniel Stenberg [Mon, 16 Feb 2004 16:27:18 +0000 (16:27 +0000)]
only build adig and ahost if 'make demos' is used

20 years agosupport closesocket() for closing sockets as well, as then we can use this
Daniel Stenberg [Mon, 16 Feb 2004 16:24:01 +0000 (16:24 +0000)]
support closesocket() for closing sockets as well, as then we can use this
code fine on ares!

20 years agoMake realloc() support NULL as pointer. Made to allow us to use these routines
Daniel Stenberg [Mon, 16 Feb 2004 16:23:19 +0000 (16:23 +0000)]
Make realloc() support NULL as pointer. Made to allow us to use these routines
to memdebug the ares stuff as well.

20 years agoitem 24 fixed, edited a few issues
Daniel Stenberg [Mon, 16 Feb 2004 15:27:40 +0000 (15:27 +0000)]
item 24 fixed, edited a few issues

20 years agoMake the 'areschannel' get created in the curl_easy_init() and re-use that
Daniel Stenberg [Mon, 16 Feb 2004 15:24:22 +0000 (15:24 +0000)]
Make the 'areschannel' get created in the curl_easy_init() and re-use that
same channel during the whole curl handle's life until curl_easy_cleanup().

20 years agoupdates
Daniel Stenberg [Mon, 16 Feb 2004 13:36:08 +0000 (13:36 +0000)]
updates

20 years agoverbose-fix, socks5-fix, dnscache-fix, configure-winmmlib-fix
Daniel Stenberg [Mon, 16 Feb 2004 13:33:41 +0000 (13:33 +0000)]
verbose-fix, socks5-fix, dnscache-fix, configure-winmmlib-fix

20 years agoFix verbosconnect() when ipv6-enabled to not assume that conn->serv_addr
Daniel Stenberg [Mon, 16 Feb 2004 13:14:55 +0000 (13:14 +0000)]
Fix verbosconnect() when ipv6-enabled to not assume that conn->serv_addr
is a valid pointer, but instead always depend on the passed-in dns pointer.
This happens to be NULL when the connection is re-used...

20 years agoremoved usage of a silly macro instead of the actual functions memcpy
Daniel Stenberg [Mon, 16 Feb 2004 09:56:18 +0000 (09:56 +0000)]
removed usage of a silly macro instead of the actual functions memcpy
and memset

20 years agoJeff Lawson pointed out that we need to check for a '5' in the version field
Daniel Stenberg [Mon, 16 Feb 2004 07:33:30 +0000 (07:33 +0000)]
Jeff Lawson pointed out that we need to check for a '5' in the version field
to properly work with SOCKS5 proxies. I also included some ascii art describing
the SOCKS5 response, as RFC1928 describes. Jeff provided details in bug
report #741841 and here: http://curl.haxx.se/mail/lib-2004-02/0181.html