platform/upstream/curl.git
16 years agoMohun Biswas pointed out that --libcurl generated a source code with an int
Daniel Stenberg [Tue, 1 Jan 2008 21:11:26 +0000 (21:11 +0000)]
Mohun Biswas pointed out that --libcurl generated a source code with an int
function but without a return statement. While fixing that, I also took care
about adding some better comments for the generated code.

16 years ago--libcurl was added in 7.16.1, a useful information
Daniel Stenberg [Thu, 27 Dec 2007 21:44:21 +0000 (21:44 +0000)]
--libcurl was added in 7.16.1, a useful information

16 years agoDmitry Kurochkin mentioned a flaw
Daniel Stenberg [Wed, 26 Dec 2007 23:29:35 +0000 (23:29 +0000)]
Dmitry Kurochkin mentioned a flaw
(http://curl.haxx.se/mail/lib-2007-12/0252.html) in detect_proxy() which
failed to set the bits.proxy variable properly when an environment variable
told libcurl to use a http proxy.

16 years agoIn an attempt to repeat the problem in bug report #1850730
Daniel Stenberg [Wed, 26 Dec 2007 21:48:52 +0000 (21:48 +0000)]
In an attempt to repeat the problem in bug report #1850730
(http://curl.haxx.se/bug/view.cgi?id=1850730) I wrote up test case 552. The
test is doing a 70K POST with a read callback and an ioctl callback over a
proxy requiring Digest auth. The test case code is more or less identical to
the test recipe code provided by Spacen Jasset (who submitted the bug report).

16 years agowhat we're having atm
Daniel Stenberg [Wed, 26 Dec 2007 21:46:51 +0000 (21:46 +0000)]
what we're having atm

16 years agoadded missing semicolon fromn last commit.
Gunter Knauf [Tue, 25 Dec 2007 13:26:01 +0000 (13:26 +0000)]
added missing semicolon fromn last commit.

16 years agoGary Maxwell filed bug report #1856628
Daniel Stenberg [Mon, 24 Dec 2007 23:45:48 +0000 (23:45 +0000)]
Gary Maxwell filed bug report #1856628
(http://curl.haxx.se/bug/view.cgi?id=1856628) and provided a fix for the
(small) memory leak in the SSL session ID caching code. It happened when a
previous entry in the cache was re-used.

16 years agoUse getcwd() to get the directory, which works even if one of the directory
Dan Fandrich [Sat, 22 Dec 2007 18:25:43 +0000 (18:25 +0000)]
Use getcwd() to get the directory, which works even if one of the directory
components doesn't have read permission set.

16 years agoUse getcwd() to get the directory, which works even if one of the
Dan Fandrich [Thu, 20 Dec 2007 21:21:43 +0000 (21:21 +0000)]
Use getcwd() to get the directory, which works even if one of the
directory components doesn't have read permission set.

16 years agoEnsure that nroff doesn't put anything but ASCII characters into the
Dan Fandrich [Wed, 19 Dec 2007 21:19:01 +0000 (21:19 +0000)]
Ensure that nroff doesn't put anything but ASCII characters into the
--manual text.

16 years ago(http://curl.haxx.se/mail/archive-2007-12/0039.html) reported and fixed
Yang Tse [Tue, 18 Dec 2007 18:33:24 +0000 (18:33 +0000)]
(curl.haxx.se/mail/archive-2007-12/0039.html) reported and fixed
a file truncation problem on Windows build targets triggered when retrying
a download with curl.

16 years agoMSVC 9.0 (VS2008) does not support Windows build targets prior to WinXP,
Yang Tse [Tue, 18 Dec 2007 18:08:19 +0000 (18:08 +0000)]
MSVC 9.0 (VS2008) does not support Windows build targets prior to WinXP,
and makes wrong asumptions of build target when it isn't specified. So,
if no build target has been defined we will target WinXP when building
with MSVC 9.0 (VS2008).

16 years agopollfd struct and WSA_poll fixes for Windows Vista already present in CVS
Yang Tse [Tue, 18 Dec 2007 10:36:32 +0000 (10:36 +0000)]
pollfd struct and WSA_poll fixes for Windows Vista already present in CVS

16 years agoMateusz Loskot pointed out that VC++ 9.0 (2008) has the pollfd struct and
Daniel Stenberg [Mon, 17 Dec 2007 21:19:42 +0000 (21:19 +0000)]
Mateusz Loskot pointed out that VC++ 9.0 (2008) has the pollfd struct and
defines in the SDK somehow differently so we have to add a define to the
config-win32.h file to make select.h compile nicely.

16 years agospell!
Daniel Stenberg [Sat, 15 Dec 2007 22:19:08 +0000 (22:19 +0000)]
spell!

16 years agoAdd test 551 that tests callback-post over a proxy that requires Digest auth.
Daniel Stenberg [Sat, 15 Dec 2007 22:13:07 +0000 (22:13 +0000)]
Add test 551 that tests callback-post over a proxy that requires Digest auth.
A failed attempt to repeat bug report #1850730 (ie the test works fine).

16 years agoremove mistaken "-d" from here
Daniel Stenberg [Fri, 14 Dec 2007 22:09:15 +0000 (22:09 +0000)]
remove mistaken "-d" from here

16 years ago-u addition: If you just give the user name (without entering a colon) curl
Daniel Stenberg [Fri, 14 Dec 2007 11:19:56 +0000 (11:19 +0000)]
-u addition: If you just give the user name (without entering a colon) curl
will prompt for a password. Denis Bredelet pointed out!

16 years agoAdded missing <features>
Dan Fandrich [Fri, 14 Dec 2007 01:09:45 +0000 (01:09 +0000)]
Added missing <features>

16 years agoFixed typo in test title
Dan Fandrich [Fri, 14 Dec 2007 01:05:30 +0000 (01:05 +0000)]
Fixed typo in test title

16 years agoFix compiler warning
Yang Tse [Thu, 13 Dec 2007 14:39:51 +0000 (14:39 +0000)]
Fix compiler warning

16 years agoDavid Wright filed bug report #1849764
Daniel Stenberg [Thu, 13 Dec 2007 10:00:06 +0000 (10:00 +0000)]
David Wright filed bug report #1849764
(http://curl.haxx.se/bug/view.cgi?id=1849764) with an included fix. He
identified a problem for re-used connections that previously had sent
Expect: 100-continue and in some situations the subsequent POST (that didn't
use Expect:) still had the internal flag set for its use. David's fix (that
makes the setting of the flag in every single request unconditionally) is
fine and is now used!

16 years agoGilles Blanc made the curl tool enable SO_KEEPALIVE for the connections and
Daniel Stenberg [Wed, 12 Dec 2007 11:22:15 +0000 (11:22 +0000)]
Gilles Blanc made the curl tool enable SO_KEEPALIVE for the connections and
added the --no-keep-alive option that can disable that on demand.

16 years agoclarify that the CURLMOPT_TIMERFUNCTION callback can pass in 0 and -1 as legal
Daniel Stenberg [Tue, 11 Dec 2007 21:19:38 +0000 (21:19 +0000)]
clarify that the CURLMOPT_TIMERFUNCTION callback can pass in 0 and -1 as legal
values and what they mean

16 years agobuild acountry too
Daniel Stenberg [Tue, 11 Dec 2007 19:34:31 +0000 (19:34 +0000)]
build acountry too

16 years agoAdded acountry.c.
Gisle Vanem [Tue, 11 Dec 2007 17:26:07 +0000 (17:26 +0000)]
Added acountry.c.

16 years agoAdded build of acountry.nlm.
Gisle Vanem [Tue, 11 Dec 2007 17:24:43 +0000 (17:24 +0000)]
Added build of acountry.nlm.

16 years agoAdded build of acountry.exe.
Gisle Vanem [Tue, 11 Dec 2007 17:23:18 +0000 (17:23 +0000)]
Added build of acountry.exe.

16 years agoBuild acountry.exe. Added 'socklen_t' define.
Gisle Vanem [Tue, 11 Dec 2007 17:22:20 +0000 (17:22 +0000)]
Build acountry.exe. Added 'socklen_t' define.

16 years agoAnother sample application that returns country-code and
Gisle Vanem [Tue, 11 Dec 2007 17:21:12 +0000 (17:21 +0000)]
Another sample application that returns country-code and
name from an IPv4-address or host-name. Using the service of
countries.nerd.dk.

16 years agogrrr, the previous commit was meant to properly make sure that we don't
Daniel Stenberg [Mon, 10 Dec 2007 22:20:26 +0000 (22:20 +0000)]
grrr, the previous commit was meant to properly make sure that we don't
link any executables when doing debug builds since they kind of assume
symbols provided by libcurl, but it also wrongly included acountry.c

16 years agowhen building
Daniel Stenberg [Mon, 10 Dec 2007 22:19:06 +0000 (22:19 +0000)]
when building

16 years agobuild ahost and adig by default but don't install them
Daniel Stenberg [Mon, 10 Dec 2007 21:42:04 +0000 (21:42 +0000)]
build ahost and adig by default but don't install them

16 years agoDefine new options in OS400 RPG interface
Patrick Monnerat [Mon, 10 Dec 2007 17:09:09 +0000 (17:09 +0000)]
Define new options in OS400 RPG interface
Port OS400 compilation scripts to >= V5R2M0

16 years agoFix for targets that do have 'struct in6_addr', but which doesn't
Gisle Vanem [Mon, 10 Dec 2007 16:14:02 +0000 (16:14 +0000)]
Fix for targets that do have 'struct in6_addr', but which doesn't
define 's6_addr' as a macro.

16 years agocut out the number of contributors from this file since it'll always be wrong
Daniel Stenberg [Mon, 10 Dec 2007 11:33:46 +0000 (11:33 +0000)]
cut out the number of contributors from this file since it'll always be wrong

16 years ago5.13 How do I stop an ongoing transfer?
Daniel Stenberg [Mon, 10 Dec 2007 10:28:56 +0000 (10:28 +0000)]
5.13 How do I stop an ongoing transfer?

16 years agoAndrew Moise filed bug report #1847501
Daniel Stenberg [Sun, 9 Dec 2007 22:31:53 +0000 (22:31 +0000)]
Andrew Moise filed bug report #1847501
(http://curl.haxx.se/bug/view.cgi?id=1847501) and pointed out a memcpy()
that should be memmove() in the convert_lineends() function.

16 years agoadd in toc too
Daniel Stenberg [Sun, 9 Dec 2007 12:26:05 +0000 (12:26 +0000)]
add in toc too

16 years agoRTMP support?
Daniel Stenberg [Sun, 9 Dec 2007 12:22:22 +0000 (12:22 +0000)]
RTMP support?

16 years agooops another bad numbering
Daniel Stenberg [Sun, 9 Dec 2007 12:20:06 +0000 (12:20 +0000)]
oops another bad numbering

16 years agooops duplicate numbering
Daniel Stenberg [Sun, 9 Dec 2007 12:12:52 +0000 (12:12 +0000)]
oops duplicate numbering

16 years agoslightly rephrased
Daniel Stenberg [Sun, 9 Dec 2007 12:00:54 +0000 (12:00 +0000)]
slightly rephrased

16 years agoRemoved use of '..\lib\libcurl_wc.lib' as this is not really
Gisle Vanem [Sun, 9 Dec 2007 09:58:56 +0000 (09:58 +0000)]
Removed use of '..\lib\libcurl_wc.lib' as this is not really
a static-lib. Renamed 'OBJ_DIR' to 'WC_Win32.obj'.

16 years agoRemoved building 'libcurl_wc.lib' as this isn't a static-library
Gisle Vanem [Sun, 9 Dec 2007 09:44:05 +0000 (09:44 +0000)]
Removed building 'libcurl_wc.lib' as this isn't a static-library
in the common sense. Renamed 'OBJ_DIR' to 'WC_Win32.obj'.

16 years agoTravelling some 500km by train back and forth on the same day gives you time
Daniel Stenberg [Sat, 8 Dec 2007 23:01:46 +0000 (23:01 +0000)]
Travelling some 500km by train back and forth on the same day gives you time
to do things you don't otherwise do, but here's the summary of today's work...

16 years agoreformat to FAQ/CONTRIBUTE style, for nicer web-look when I apply the magic
Daniel Stenberg [Sat, 8 Dec 2007 23:00:00 +0000 (23:00 +0000)]
reformat to FAQ/CONTRIBUTE style, for nicer web-look when I apply the magic
script(s) on it online

16 years agocleanup
Daniel Stenberg [Sat, 8 Dec 2007 22:58:12 +0000 (22:58 +0000)]
cleanup

16 years agofix a crash in oom situations (thanks runtests.pl -t!)
Daniel Stenberg [Sat, 8 Dec 2007 22:57:17 +0000 (22:57 +0000)]
fix a crash in oom situations (thanks runtests.pl -t!)

16 years agoadd keywords
Daniel Stenberg [Sat, 8 Dec 2007 22:56:17 +0000 (22:56 +0000)]
add keywords

16 years agoadd missing files
Daniel Stenberg [Sat, 8 Dec 2007 22:56:05 +0000 (22:56 +0000)]
add missing files

16 years agocorrect the comment about size
Daniel Stenberg [Sat, 8 Dec 2007 22:53:49 +0000 (22:53 +0000)]
correct the comment about size

16 years agoadd test 549 and 550
Daniel Stenberg [Sat, 8 Dec 2007 22:53:28 +0000 (22:53 +0000)]
add test 549 and 550

16 years agomention how to enable chunked encoding for POSTs
Daniel Stenberg [Sat, 8 Dec 2007 22:52:39 +0000 (22:52 +0000)]
mention how to enable chunked encoding for POSTs

16 years agoAll static functions that were previously name Curl_* something no longer
Daniel Stenberg [Sat, 8 Dec 2007 22:50:55 +0000 (22:50 +0000)]
All static functions that were previously name Curl_* something no longer
use that prefix as we use that prefix only for library-wide internal global
symbols.

16 years agoclarify that when curl_multi_timeout() returns -1 it just means that there
Daniel Stenberg [Thu, 6 Dec 2007 22:36:52 +0000 (22:36 +0000)]
clarify that when curl_multi_timeout() returns -1 it just means that there
is no current timeout. It does not mean wait forever and it does not mean
do not wait at all. It means there is no timeout value known at this point in
time.

16 years agoSpacen Jasset reported a problem with doing POST (with data read with a
Daniel Stenberg [Wed, 5 Dec 2007 21:20:14 +0000 (21:20 +0000)]
Spacen Jasset reported a problem with doing POST (with data read with a
callback) over a proxy when NTLM is used as auth with the proxy. The bug
also concerned Digest and was limited to using callback only. Spacen worked
with us to provide a useful patch. I added the test case 547 and 548 to
verify two variations of POST over proxy with NTLM.

16 years agofix compiler warning
Daniel Stenberg [Wed, 5 Dec 2007 11:10:24 +0000 (11:10 +0000)]
fix compiler warning

16 years agoadded test548 which uses the lib547 source file, preparing for test547 which
Daniel Stenberg [Wed, 5 Dec 2007 11:08:56 +0000 (11:08 +0000)]
added test548 which uses the lib547 source file, preparing for test547 which
is supposed to repeat the bug report "NTLM proxy authentication with
CURLOPT_READDATA seems broken." posted on the curl-library mailing list on dec
3 2007.

16 years agoFix compiler warning: variable may be used uninitialized
Yang Tse [Tue, 4 Dec 2007 00:15:03 +0000 (00:15 +0000)]
Fix compiler warning: variable may be used uninitialized

16 years agoRay Pekowski filed bug report #1842029
Daniel Stenberg [Mon, 3 Dec 2007 22:44:47 +0000 (22:44 +0000)]
Ray Pekowski filed bug report #1842029

16 years agoFix three issues previous cleanup introduces.
Yang Tse [Mon, 3 Dec 2007 19:57:18 +0000 (19:57 +0000)]
Fix three issues previous cleanup introduces.

16 years agoSSL session id caching bugfix
Daniel Stenberg [Mon, 3 Dec 2007 11:49:20 +0000 (11:49 +0000)]
SSL session id caching bugfix

16 years agoBug report #1842029 (http://curl.haxx.se/bug/view.cgi?id=1842029) identified
Daniel Stenberg [Mon, 3 Dec 2007 11:48:09 +0000 (11:48 +0000)]
Bug report #1842029 (curl.haxx.se/bug/view.cgi?id=1842029) identified
a problem with SSL session caching that prevent it from working, and the
associated fix!

16 years agomention "no longer default-appends ;type= on FTP URLs thru proxies" as a bug
Daniel Stenberg [Mon, 3 Dec 2007 11:41:36 +0000 (11:41 +0000)]
mention "no longer default-appends ;type= on FTP URLs thru proxies" as a bug
fix even if kind of implied by the new option

16 years agoNow libcurl (built with OpenSSL) doesn't return error anymore if the remote
Daniel Stenberg [Mon, 3 Dec 2007 11:39:27 +0000 (11:39 +0000)]
Now libcurl (built with OpenSSL) doesn't return error anymore if the remote
SSL-based server doesn't present a certificate when the request is told to
ignore certificate verification anyway.

16 years ago Erik Kline cleaned up ares_gethostbyaddr.c:next_lookup() somewhat
Daniel Stenberg [Mon, 3 Dec 2007 10:25:05 +0000 (10:25 +0000)]
 Erik Kline cleaned up ares_gethostbyaddr.c:next_lookup() somewhat

16 years agoBrad Spencer fixed the configure script to assume that there's no
Daniel Stenberg [Mon, 3 Dec 2007 10:22:29 +0000 (10:22 +0000)]
Brad Spencer fixed the configure script to assume that there's no
/dev/urandom when built cross-compiled as then the script cannot check for
it.

16 years agoremoved the ;type= thing for FTP urls through proxy, since that's now only
Daniel Stenberg [Mon, 3 Dec 2007 09:50:32 +0000 (09:50 +0000)]
removed the ;type= thing for FTP urls through proxy, since that's now only
present when enabled by on option which isn't done by default (and isn't even
available for the curl app atm)

16 years ago107 - resolve the type= thing for FTP URLs over HTTP proxies, is solved
Daniel Stenberg [Sun, 2 Dec 2007 23:39:39 +0000 (23:39 +0000)]
107 - resolve the type= thing for FTP URLs over HTTP proxies, is solved

16 years agoMichal Marek introduced CURLOPT_PROXY_TRANSFER_MODE which is used to control
Daniel Stenberg [Sun, 2 Dec 2007 23:38:23 +0000 (23:38 +0000)]
Michal Marek introduced CURLOPT_PROXY_TRANSFER_MODE which is used to control
the appending of the "type=" thing on FTP URLs when they are passed to a
HTTP proxy. Some proxies just don't like that appending (which is done
unconditionally in 7.17.1), and some proxies treat binary/ascii transfers
better with the appending done!

16 years agoUpped copyright year
Dan Fandrich [Fri, 30 Nov 2007 02:31:07 +0000 (02:31 +0000)]
Upped copyright year

16 years agouh, corrected pretty major write error!
Daniel Stenberg [Thu, 29 Nov 2007 22:27:51 +0000 (22:27 +0000)]
uh, corrected pretty major write error!

16 years agoftp resumed upload and long Digest nonces
Daniel Stenberg [Thu, 29 Nov 2007 22:15:22 +0000 (22:15 +0000)]
ftp resumed upload and long Digest nonces

16 years agoA bug report on the curl-library list showed a HTTP Digest session going on
Daniel Stenberg [Thu, 29 Nov 2007 22:14:48 +0000 (22:14 +0000)]
A bug report on the curl-library list showed a HTTP Digest session going on
with a 700+ letter nonce. Previously libcurl only support 127 letter ones
and now I bumped it to 1023.

16 years agoFixed the resumed FTP upload loop to not require that the read callback
Daniel Stenberg [Thu, 29 Nov 2007 22:14:33 +0000 (22:14 +0000)]
Fixed the resumed FTP upload loop to not require that the read callback
returns a full buffer on each invoke.

16 years agoinclude the libssh2 return code in the output for these failures to ease
Daniel Stenberg [Thu, 29 Nov 2007 11:25:10 +0000 (11:25 +0000)]
include the libssh2 return code in the output for these failures to ease
debugging

16 years agothe gethostbyname fix applied here as well
Daniel Stenberg [Wed, 28 Nov 2007 15:18:27 +0000 (15:18 +0000)]
the gethostbyname fix applied here as well

16 years agofix next_lookup() to continue searching even if c-ares failed to load the
Daniel Stenberg [Wed, 28 Nov 2007 10:46:40 +0000 (10:46 +0000)]
fix next_lookup() to continue searching even if c-ares failed to load the
/etc/hosts file, pointed out by Erik Kline:
http://daniel.haxx.se/projects/c-ares/mail/c-ares-archive-2007-11/0027.shtml

16 years agoWhen --with-gssapi (without given path) is used, we must use krb5-config to
Daniel Stenberg [Wed, 28 Nov 2007 10:33:47 +0000 (10:33 +0000)]
When --with-gssapi (without given path) is used, we must use krb5-config to
get the libs as well and not only the include path like we used to.

16 years agoTo allow remote log inspection avoid redirecting messages to stderr.
Yang Tse [Wed, 28 Nov 2007 01:46:28 +0000 (01:46 +0000)]
To allow remote log inspection avoid redirecting messages to stderr.
Cleanup some debugging messages. Unlink log file on exit.

16 years agoRemove the check for libdl since that isn't actually used and it causes
Daniel Stenberg [Tue, 27 Nov 2007 22:41:53 +0000 (22:41 +0000)]
Remove the check for libdl since that isn't actually used and it causes
warnings. Pointed out by Robin Cornelius.

16 years agopkgconfig fix by Andreas Schuldei
Daniel Stenberg [Tue, 27 Nov 2007 22:38:11 +0000 (22:38 +0000)]
pkgconfig fix by Andreas Schuldei

16 years agospellfix
Daniel Stenberg [Tue, 27 Nov 2007 22:37:55 +0000 (22:37 +0000)]
spellfix

16 years agoConnectTimeout requires OpenSSH 3.7 or later
Yang Tse [Tue, 27 Nov 2007 20:57:22 +0000 (20:57 +0000)]
ConnectTimeout requires OpenSSH 3.7 or later

16 years agoExplicitly disallow remote hosts to connect to local forwarded ports,
Yang Tse [Tue, 27 Nov 2007 00:52:30 +0000 (00:52 +0000)]
Explicitly disallow remote hosts to connect to local forwarded ports,
the socks server port in the test suite. This is the default setting
unless a tinkered built ssh is being used.

16 years agoStop ssh and socks servers when verification fails
Yang Tse [Mon, 26 Nov 2007 14:26:40 +0000 (14:26 +0000)]
Stop ssh and socks servers when verification fails

16 years agoProviding an explicit bind address besides the port for dynamic application-level
Yang Tse [Mon, 26 Nov 2007 14:07:09 +0000 (14:07 +0000)]
Providing an explicit bind address besides the port for dynamic application-level
port forwarding, our socks port, prevents ssh from running on some systems.

By default, ssh binds local port forwardings to the loopback address, since this
was the address being given as the explicit bind address, now it isn't given.

16 years agomore blurb
Daniel Stenberg [Mon, 26 Nov 2007 12:26:58 +0000 (12:26 +0000)]
more blurb

16 years agoAdded recent changes and spellchecked
Daniel Stenberg [Mon, 26 Nov 2007 11:04:51 +0000 (11:04 +0000)]
Added recent changes and spellchecked

16 years agotest1015 --data-urlencode
Daniel Stenberg [Mon, 26 Nov 2007 11:04:21 +0000 (11:04 +0000)]
test1015 --data-urlencode

16 years ago#1 fixed --data-urlencode when no = or @ was used
Daniel Stenberg [Mon, 26 Nov 2007 11:03:32 +0000 (11:03 +0000)]
#1 fixed --data-urlencode when no = or @ was used
#2 extended the user-agent buffer since I hit the 128 byte boundary!

16 years agoslightly less outdated
Daniel Stenberg [Mon, 26 Nov 2007 11:02:45 +0000 (11:02 +0000)]
slightly less outdated

16 years agoTemporary change to better debug startup failures
Yang Tse [Mon, 26 Nov 2007 02:45:24 +0000 (02:45 +0000)]
Temporary change to better debug startup failures
of test suite ssh and socks servers.

16 years agoAllow different start timeout specification for each server
Yang Tse [Sun, 25 Nov 2007 03:55:53 +0000 (03:55 +0000)]
Allow different start timeout specification for each server

16 years agoreqdata doesn't exist anymore and the path moved to the UrlState struct
Daniel Stenberg [Sat, 24 Nov 2007 23:18:21 +0000 (23:18 +0000)]
reqdata doesn't exist anymore and the path moved to the UrlState struct

16 years agostruct HandleData is now called struct SingleRequest, and is only for data that
Daniel Stenberg [Sat, 24 Nov 2007 23:16:55 +0000 (23:16 +0000)]
struct HandleData is now called struct SingleRequest, and is only for data that
is inited at the start of the DO action. I removed the Curl_transfer_keeper
struct completely, and I had to move out a few struct members (that had to
be set before DO or used after DONE) to the UrlState struct. The SingleRequest
struct is accessed with SessionHandle->req.

One of the biggest reasons for doing this was the bunch of duplicate struct
members in HandleData and Curl_transfer_keeper since it was really messy to
keep track of two variables with the same name and basically the same purpose!

16 years agomake 'checkdied' in runtests.pl more robust
Yang Tse [Fri, 23 Nov 2007 12:18:45 +0000 (12:18 +0000)]
make 'checkdied' in runtests.pl more robust

16 years agoRevert last change since it breaks running the test suite
Yang Tse [Fri, 23 Nov 2007 09:50:44 +0000 (09:50 +0000)]
Revert last change since it breaks running the test suite
when builddir is different from srcdir.

16 years agoImprove chance of running runtests.pl from outside the
Yang Tse [Fri, 23 Nov 2007 04:03:46 +0000 (04:03 +0000)]
Improve chance of running runtests.pl from outside the
source tree 'tests' directory