platform/upstream/curl.git
18 years agoGet qualifier of arg 1 for getnameinfo apart. Take 3.
Yang Tse [Sat, 1 Jul 2006 11:21:38 +0000 (11:21 +0000)]
Get qualifier of arg 1 for getnameinfo apart. Take 3.

18 years agoGet qualifier of arg 1 for getnameinfo apart.
Yang Tse [Sat, 1 Jul 2006 03:07:07 +0000 (03:07 +0000)]
Get qualifier of arg 1 for getnameinfo apart.

18 years agoGet qualifier of arg 1 for getnameinfo apart. Take 2.
Yang Tse [Sat, 1 Jul 2006 02:53:18 +0000 (02:53 +0000)]
Get qualifier of arg 1 for getnameinfo apart. Take 2.

18 years agoGet qualifier of arg 1 for getnameinfo apart.
Yang Tse [Fri, 30 Jun 2006 19:20:16 +0000 (19:20 +0000)]
Get qualifier of arg 1 for getnameinfo apart.

18 years agotypecast the number passed to CURLOPT_INFILESIZE_LARGE as a curl_off_t
Daniel Stenberg [Fri, 30 Jun 2006 10:26:26 +0000 (10:26 +0000)]
typecast the number passed to CURLOPT_INFILESIZE_LARGE as a curl_off_t

18 years agoRemove experimental notice from CURL_CHECK_FUNC_GETNAMEINFO
Yang Tse [Fri, 30 Jun 2006 00:22:38 +0000 (00:22 +0000)]
Remove experimental notice from CURL_CHECK_FUNC_GETNAMEINFO

18 years agowith a very recent yassl, we now can display 'yassl' when the OpenSSL API is
Daniel Stenberg [Thu, 29 Jun 2006 07:35:02 +0000 (07:35 +0000)]
with a very recent yassl, we now can display 'yassl' when the OpenSSL API is
in fact provided by yassl instead

18 years agochanged wording on the curl_multi_fdset() problem and moved the -K change from
Daniel Stenberg [Wed, 28 Jun 2006 11:31:50 +0000 (11:31 +0000)]
changed wording on the curl_multi_fdset() problem and moved the -K change from
bugfixes

18 years agofix better minor compiler warning
Yang Tse [Wed, 28 Jun 2006 05:22:47 +0000 (05:22 +0000)]
fix better minor compiler warning

18 years agofix minor compiler warning
Yang Tse [Wed, 28 Jun 2006 04:17:04 +0000 (04:17 +0000)]
fix minor compiler warning

18 years agofix minor compiler warning
Yang Tse [Wed, 28 Jun 2006 02:45:27 +0000 (02:45 +0000)]
fix minor compiler warning

18 years agoinclude config-win32.h in release archives
Daniel Stenberg [Mon, 26 Jun 2006 08:56:04 +0000 (08:56 +0000)]
include config-win32.h in release archives

18 years agochanged wording slightly, and added standard boiler-plate header
Daniel Stenberg [Mon, 26 Jun 2006 06:43:45 +0000 (06:43 +0000)]
changed wording slightly, and added standard boiler-plate header

18 years agoWallner's update
Daniel Stenberg [Sat, 24 Jun 2006 23:11:36 +0000 (23:11 +0000)]
Wallner's update

18 years agoAdded README.pipelining but also extracted the "docs" files to a separate list
Daniel Stenberg [Sat, 24 Jun 2006 21:54:22 +0000 (21:54 +0000)]
Added README.pipelining but also extracted the "docs" files to a separate list
to get a better overview

18 years agothoughts and ideas as posted to the list the other day
Daniel Stenberg [Sat, 24 Jun 2006 21:51:28 +0000 (21:51 +0000)]
thoughts and ideas as posted to the list the other day

18 years agocorrected introduction version number
Daniel Stenberg [Sat, 24 Jun 2006 21:49:40 +0000 (21:49 +0000)]
corrected introduction version number

18 years agoMichael Wallner added curl_formget(), which allows an application to extract
Daniel Stenberg [Sat, 24 Jun 2006 21:46:41 +0000 (21:46 +0000)]
Michael Wallner added curl_formget(), which allows an application to extract
(serialise) a previously built formpost (as with curl_formadd()).

18 years ago1.3.1
Daniel Stenberg [Sat, 24 Jun 2006 18:29:18 +0000 (18:29 +0000)]
1.3.1

18 years agoProvide a libcurl API for setting mutex callbacks in the underlying SSL
Daniel Stenberg [Sat, 24 Jun 2006 15:21:49 +0000 (15:21 +0000)]
Provide a libcurl API for setting mutex callbacks in the underlying SSL
library, so that the same application code can use mutex-locking
independently of OpenSSL or GnutTLS being used.

18 years agoArve Knudsen found a flaw in curl_multi_fdset() for systems where
Daniel Stenberg [Fri, 23 Jun 2006 22:07:06 +0000 (22:07 +0000)]
Arve Knudsen found a flaw in curl_multi_fdset() for systems where
curl_socket_t is unsigned (like Windows) that could cause it to wrongly
return a max fd of -1.

18 years agoPeter Silva introduced CURLOPT_MAX_SEND_SPEED_LARGE and
Daniel Stenberg [Thu, 22 Jun 2006 21:36:53 +0000 (21:36 +0000)]
Peter Silva introduced CURLOPT_MAX_SEND_SPEED_LARGE and
CURLOPT_MAX_RECV_SPEED_LARGE that limit tha maximum rate libcurl is allowed
to send or receive data. This kind of adds the the command line tool's
option --limit-rate to the library.

The rate limiting logic in the curl app is now removed and is instead
provided by libcurl itself. Transfer rate limiting will now also work for -d
and -F, which it didn't before.

18 years agominor language edits bug reports 1510080 1510098
Daniel Stenberg [Wed, 21 Jun 2006 17:34:29 +0000 (17:34 +0000)]
minor language edits bug reports 1510080 1510098

18 years agothe D binding link is dead but we know of no new one!
Daniel Stenberg [Tue, 20 Jun 2006 07:27:10 +0000 (07:27 +0000)]
the D binding link is dead but we know of no new one!

18 years agobad syntax
Daniel Stenberg [Tue, 20 Jun 2006 07:03:29 +0000 (07:03 +0000)]
bad syntax

18 years agomake -K on a bad file now displays a warning
Daniel Stenberg [Mon, 19 Jun 2006 21:39:57 +0000 (21:39 +0000)]
make -K on a bad file now displays a warning

18 years agoRemove "big endian" DNS section and RR data integer parser macros from
William Ahern [Mon, 19 Jun 2006 06:41:55 +0000 (06:41 +0000)]
Remove "big endian" DNS section and RR data integer parser macros from
ares_dns.h, which break c-ares on my Sparc64. Bit-wise operations in C
operate on logical values. And in any event the octets are already in
big-endian (aka network) byte order so they're being reversed (thus the
source of the breakage).

18 years agoHandle EAGAIN/EWOULDBLOCK readiness errors, which can occur for both TCP and
William Ahern [Mon, 19 Jun 2006 01:18:05 +0000 (01:18 +0000)]
Handle EAGAIN/EWOULDBLOCK readiness errors, which can occur for both TCP and
UDP even when a poll(2) or select(2) suggest otherwise.

18 years agowhen mentioning the default config file, point back to the actual description
Daniel Stenberg [Fri, 16 Jun 2006 07:27:06 +0000 (07:27 +0000)]
when mentioning the default config file, point back to the actual description
of how to write such a file

18 years agoselect_res is not a socket, it should be a plain int
Daniel Stenberg [Thu, 15 Jun 2006 21:30:32 +0000 (21:30 +0000)]
select_res is not a socket, it should be a plain int

18 years agoCheck whether gcc supports --enable-hidden-symbols before allowing it.
Dan Fandrich [Tue, 13 Jun 2006 17:43:00 +0000 (17:43 +0000)]
Check whether gcc supports --enable-hidden-symbols before allowing it.

18 years agoImplemented --enable-hidden-symbols configure option to enable
Dan Fandrich [Mon, 12 Jun 2006 20:33:04 +0000 (20:33 +0000)]
Implemented --enable-hidden-symbols configure option to enable
-fvisibility=hidden on gcc >= 4.0.  This reduces the size of the libcurl
binary and speeds up dynamic linking by hiding all the internal symbols from
the symbol table.

18 years agooops
Daniel Stenberg [Mon, 12 Jun 2006 09:32:39 +0000 (09:32 +0000)]
oops

18 years agoadded contributors from the 7.15.4 release
Daniel Stenberg [Mon, 12 Jun 2006 09:30:14 +0000 (09:30 +0000)]
added contributors from the 7.15.4 release

18 years agostarting the journey towards the next release
Daniel Stenberg [Mon, 12 Jun 2006 07:24:14 +0000 (07:24 +0000)]
starting the journey towards the next release

18 years ago7.15.4 coming up
Daniel Stenberg [Mon, 12 Jun 2006 06:53:10 +0000 (06:53 +0000)]
7.15.4 coming up

18 years agoNah, we refer people to the generic site that lists the mirrors instead of
Daniel Stenberg [Mon, 12 Jun 2006 06:51:15 +0000 (06:51 +0000)]
Nah, we refer people to the generic site that lists the mirrors instead of
trying to maintain an accurate list in this file (too).

18 years agoFix "'x' might be used uninitialized in this function" warnings.
Gisle Vanem [Sat, 10 Jun 2006 17:35:28 +0000 (17:35 +0000)]
Fix "'x' might be used uninitialized in this function" warnings.

18 years agosync with existing list of up-to-date mirrors
Daniel Stenberg [Fri, 9 Jun 2006 21:08:02 +0000 (21:08 +0000)]
sync with existing list of up-to-date mirrors

18 years agoproper use of newlines
Daniel Stenberg [Fri, 9 Jun 2006 12:07:34 +0000 (12:07 +0000)]
proper use of newlines

18 years agostricter type use to please compilers
Daniel Stenberg [Fri, 9 Jun 2006 08:25:16 +0000 (08:25 +0000)]
stricter type use to please compilers

18 years agooops, serious breakage in the fdset() function
Daniel Stenberg [Fri, 9 Jun 2006 07:08:34 +0000 (07:08 +0000)]
oops, serious breakage in the fdset() function

18 years agolib525.c does a FTP upload with PORT using multi interface
Daniel Stenberg [Thu, 8 Jun 2006 22:43:21 +0000 (22:43 +0000)]
lib525.c does a FTP upload with PORT using multi interface

18 years agocorrected the CURLINFO_TOTAL_TIME description
Daniel Stenberg [Thu, 8 Jun 2006 11:06:26 +0000 (11:06 +0000)]
corrected the CURLINFO_TOTAL_TIME description

18 years agoBrian Dessent's fixes for cygwin builds
Daniel Stenberg [Thu, 8 Jun 2006 06:12:30 +0000 (06:12 +0000)]
Brian Dessent's fixes for cygwin builds

18 years agoNTLM2 session response support
Daniel Stenberg [Wed, 7 Jun 2006 14:14:04 +0000 (14:14 +0000)]
NTLM2 session response support

18 years agotwo new mirrors and a recount of them
Daniel Stenberg [Wed, 31 May 2006 05:49:37 +0000 (05:49 +0000)]
two new mirrors and a recount of them

18 years agoAdded config-tpf.h to the release package
Daniel Stenberg [Wed, 31 May 2006 05:17:27 +0000 (05:17 +0000)]
Added config-tpf.h to the release package

18 years agoRemoved a few fixed issues and a few issues currently in progress in the Hiper
Daniel Stenberg [Tue, 30 May 2006 08:45:59 +0000 (08:45 +0000)]
Removed a few fixed issues and a few issues currently in progress in the Hiper
project. Also added a few obvious ones.

18 years agocorrect explicit/implicit terms for FTPS
Daniel Stenberg [Sun, 28 May 2006 11:28:40 +0000 (11:28 +0000)]
correct explicit/implicit terms for FTPS

18 years agoone modified proto and one removed proto
Daniel Stenberg [Sat, 27 May 2006 22:26:41 +0000 (22:26 +0000)]
one modified proto and one removed proto

18 years agoadapted to the new Curl_splayremovebyaddr() proto
Daniel Stenberg [Sat, 27 May 2006 22:26:16 +0000 (22:26 +0000)]
adapted to the new Curl_splayremovebyaddr() proto

18 years agoIfdef'ed out unused function, added lots of comments and renamed a few
Daniel Stenberg [Sat, 27 May 2006 22:25:55 +0000 (22:25 +0000)]
Ifdef'ed out unused function, added lots of comments and renamed a few
variables, simplified the splayprint function, modified
Curl_splayremovebyaddr() to return error code. All in an effort to track down
the reported splay problem, but I've failed to do that so far...

18 years agoreverted the previous .libs check for libs, but instead make mingw32 builds
Daniel Stenberg [Sat, 27 May 2006 11:36:05 +0000 (11:36 +0000)]
reverted the previous .libs check for libs, but instead make mingw32 builds
more similar to other configure builds (== use .la for libext and no binext)

18 years agoÓscar Morales Vivó updated the libcurl.framework.make file.
Daniel Stenberg [Fri, 26 May 2006 22:23:54 +0000 (22:23 +0000)]
Óscar Morales Vivó updated the libcurl.framework.make file.

18 years agolong/int cleanup to silence picky compiler warnings
Daniel Stenberg [Fri, 26 May 2006 11:26:42 +0000 (11:26 +0000)]
long/int cleanup to silence picky compiler warnings

18 years agoOlaf Stüben fixed a bug that caused Digest authentication with md5-sess to
Daniel Stenberg [Thu, 25 May 2006 23:04:20 +0000 (23:04 +0000)]
Olaf Stüben fixed a bug that caused Digest authentication with md5-sess to
fail. When using the md5-sess, the result was not Md5 encoded and Base64
transformed.

18 years agominor RFC updates, Dan Fandrich brought my attention to them
Daniel Stenberg [Thu, 25 May 2006 11:15:25 +0000 (11:15 +0000)]
minor RFC updates, Dan Fandrich brought my attention to them

18 years agobetter check for libs created in the .libs directory since libtool does this
Daniel Stenberg [Thu, 25 May 2006 11:04:08 +0000 (11:04 +0000)]
better check for libs created in the .libs directory since libtool does this
kind of magic

18 years agoadded some missing items
Daniel Stenberg [Wed, 24 May 2006 23:16:22 +0000 (23:16 +0000)]
added some missing items

18 years agoCopied the NO_UNDEFINED magic from libcurl to make this build fine again with
Daniel Stenberg [Wed, 24 May 2006 23:02:51 +0000 (23:02 +0000)]
Copied the NO_UNDEFINED magic from libcurl to make this build fine again with
libtool cross-compiled on linux with mingw32

18 years agoMichael Wallner provided a patch that allows "SESS" to be set with
Daniel Stenberg [Wed, 24 May 2006 22:46:38 +0000 (22:46 +0000)]
Michael Wallner provided a patch that allows "SESS" to be set with
CURLOPT_COOKIELIST, which then makes all session cookies get cleared. (slightly
edited by me, and the re-indent in cookie.c was also done by me)

18 years agomake sure we pass a time_t * to localtime(), and the timeval struct members
Daniel Stenberg [Wed, 24 May 2006 21:39:52 +0000 (21:39 +0000)]
make sure we pass a time_t * to localtime(), and the timeval struct members
are not always time_t ones

18 years agominor fix to make Curl_splayremove() return a NULL as "removed" in case
Daniel Stenberg [Wed, 24 May 2006 16:11:31 +0000 (16:11 +0000)]
minor fix to make Curl_splayremove() return a NULL as "removed" in case
nothing matched fine

18 years agobased on Tor Arntsen's fix, this should correct test case 271 to again run
Daniel Stenberg [Wed, 24 May 2006 15:22:03 +0000 (15:22 +0000)]
based on Tor Arntsen's fix, this should correct test case 271 to again run
fine

18 years agoFixed a shell script syntax error that all of a sudden started causing this
Daniel Stenberg [Tue, 23 May 2006 22:55:46 +0000 (22:55 +0000)]
Fixed a shell script syntax error that all of a sudden started causing this
script to fail on debian unstable (some specific bash version perhaps?)

18 years agoDavid McCreedy's update
Daniel Stenberg [Tue, 23 May 2006 21:19:36 +0000 (21:19 +0000)]
David McCreedy's update

18 years agoupdated with more recent facts
Daniel Stenberg [Mon, 15 May 2006 08:09:07 +0000 (08:09 +0000)]
updated with more recent facts

18 years agoThe SOCKS connection codes don't properly acknowledge (connect) timeouts.
Daniel Stenberg [Sun, 14 May 2006 22:49:23 +0000 (22:49 +0000)]
The SOCKS connection codes don't properly acknowledge (connect) timeouts.

18 years agoThe new ftpuploadresume.c example by Philip Bock
Daniel Stenberg [Thu, 11 May 2006 22:24:44 +0000 (22:24 +0000)]
The new ftpuploadresume.c example by Philip Bock

18 years agoOk, when checking for old-style SSLeay headers we cannot just use
Daniel Stenberg [Thu, 11 May 2006 21:37:58 +0000 (21:37 +0000)]
Ok, when checking for old-style SSLeay headers we cannot just use
AC_CHECK_HEADERS() and the action-if-found since that action is run even if
just one of the six headers is found and I just now fell over a case with
a duplicate file name (a krb4 implementation with an err.h file).

I converted the check to manually make sure three of the headers are present
before considering them fine.

18 years ago1 - allow much longer time for the test FTP server to startup and get verified
Daniel Stenberg [Thu, 11 May 2006 06:34:30 +0000 (06:34 +0000)]
1 - allow much longer time for the test FTP server to startup and get verified
2 - store the time it took to verify it and allow that time to be used as
  %FTPTIME[23] in command lines to allow us to adjust better to slow hosts
  since test 190 failed on my slow solaris machine just because it hadn't
  gotten time to run all the way the test assumed all machines would reach
  before the time-out elapsed.

18 years agomake sure the LASTSOCKET check only checks for SSL status if the socket
Daniel Stenberg [Thu, 11 May 2006 05:17:40 +0000 (05:17 +0000)]
make sure the LASTSOCKET check only checks for SSL status if the socket
truly use SSL

18 years agosilence warning
Daniel Stenberg [Thu, 11 May 2006 05:16:38 +0000 (05:16 +0000)]
silence warning

18 years agoDavid McCreedy provided a fix for CURLINFO_LASTSOCKET that does extended
Daniel Stenberg [Wed, 10 May 2006 22:17:42 +0000 (22:17 +0000)]
David McCreedy provided a fix for CURLINFO_LASTSOCKET that does extended
checks on the to-be-returned socket to make sure it truly seems to be alive
and well. For SSL connection it (only) uses OpenSSL functions.

18 years agoMy Solaris test server was simply too slow to be able to respond within 4
Daniel Stenberg [Wed, 10 May 2006 21:38:46 +0000 (21:38 +0000)]
My Solaris test server was simply too slow to be able to respond within 4
seconds even when everything is fine! Now we allow a test server 8 seconds
to respond to still be considered ok.

18 years agocvsignore these files
Daniel Stenberg [Wed, 10 May 2006 14:16:30 +0000 (14:16 +0000)]
cvsignore these files

18 years ago1 - allow DICT with properly URL-escaped words, like using %20 for spaces
Daniel Stenberg [Wed, 10 May 2006 11:44:31 +0000 (11:44 +0000)]
1 - allow DICT with properly URL-escaped words, like using %20 for spaces
2 - properly escape certain letters within a DICT word to comply to the RFC2229

18 years agoremoved variable declarations shadowing previously declared variables
Daniel Stenberg [Wed, 10 May 2006 09:53:52 +0000 (09:53 +0000)]
removed variable declarations shadowing previously declared variables

18 years agoBram Matthys brought my attention to a libtool peculiarity where detecting
Daniel Stenberg [Wed, 10 May 2006 08:03:54 +0000 (08:03 +0000)]
Bram Matthys brought my attention to a libtool peculiarity where detecting
things such as C++ compiler actually is a bad thing and since we don't need
that detection I added a work-around, much inspired by a previous patch by
Paolo Bonzini. This also shortens the configure script quite a lot.

18 years agooops, could return an uninitialized variable
Daniel Stenberg [Tue, 9 May 2006 13:02:53 +0000 (13:02 +0000)]
oops, could return an uninitialized variable

18 years agoAndreas Ntaflos reported a bug in libcurl.m4: When configuring my GNU
Daniel Stenberg [Tue, 9 May 2006 12:56:35 +0000 (12:56 +0000)]
Andreas Ntaflos reported a bug in libcurl.m4: When configuring my GNU
autotools project, which optionally (default=yes) uses libcurl on a system
without a (usable) libcurl installation, but not specifying
`--without-libcurl', configure determines correctly that no libcurl is
available, however, the LIBCURL variable gets expanded to `LIBCURL = -lcurl'
in the resulting Makefiles.

David Shaw fixed the flaw.

18 years agomention the other TFTP cleanup sweep from yday
Daniel Stenberg [Tue, 9 May 2006 12:44:11 +0000 (12:44 +0000)]
mention the other TFTP cleanup sweep from yday

18 years agoRobson Braga Araujo fixed two problems in the recently added non-blocking SSL
Daniel Stenberg [Tue, 9 May 2006 12:43:49 +0000 (12:43 +0000)]
Robson Braga Araujo fixed two problems in the recently added non-blocking SSL
connects. The state machine was not reset properly so that subsequent
connects using the same handle would fail, and there were two memory leaks.

18 years agoRobson Braga Araujo fixed a memory leak when you added an easy handle to a
Daniel Stenberg [Tue, 9 May 2006 11:33:00 +0000 (11:33 +0000)]
Robson Braga Araujo fixed a memory leak when you added an easy handle to a
multi stack and that easy handle had already been used to do one or more
easy interface transfers, as then the code threw away the previously used
DNS cache without properly freeing it.

18 years agocheck more return codes and skip the initial slash in given file names
Daniel Stenberg [Mon, 8 May 2006 22:23:33 +0000 (22:23 +0000)]
check more return codes and skip the initial slash in given file names

18 years agono longer uses errno but Curl_sockerrno() and now acknowledges return codes
Daniel Stenberg [Mon, 8 May 2006 21:00:44 +0000 (21:00 +0000)]
no longer uses errno but Curl_sockerrno() and now acknowledges return codes
from Curl_client_write

18 years agoStop sending retransmitted received blocks up to client
Dan Fandrich [Mon, 8 May 2006 19:41:26 +0000 (19:41 +0000)]
Stop sending retransmitted received blocks up to client
Fixed handling of retransmitted blocks on transmit
Properly aligned data to transmit within packet
Replaced calls to strerror() with Curl_strerror()

18 years agoFixed known bug #28. The TFTP code no longer assumes a packed struct and
Daniel Stenberg [Mon, 8 May 2006 15:09:50 +0000 (15:09 +0000)]
Fixed known bug #28. The TFTP code no longer assumes a packed struct and
thus works reliably on more platforms.

18 years agoFix GnuTLS compile warning. Risking breakage with some older version of GnuTLS?
Daniel Stenberg [Sun, 7 May 2006 18:27:36 +0000 (18:27 +0000)]
Fix GnuTLS compile warning. Risking breakage with some older version of GnuTLS?

18 years agoCurl_https_getsock() was OpenSSL-specific and really should not be present
Daniel Stenberg [Fri, 5 May 2006 22:14:40 +0000 (22:14 +0000)]
Curl_https_getsock() was OpenSSL-specific and really should not be present
like this in this source file. The quickfix for now is to provide a simple
version for GnuTLS builds. The GnuTLS version of libcurl doesn't yet allow
fully non-blocking connects anyway so this function doesn't get used.

18 years agoget the Curl_sockerrno proto
Daniel Stenberg [Fri, 5 May 2006 22:07:01 +0000 (22:07 +0000)]
get the Curl_sockerrno proto

18 years agotwo more contributors
Daniel Stenberg [Fri, 5 May 2006 21:08:09 +0000 (21:08 +0000)]
two more contributors

18 years agoadditional renames of Curl_ourerrno => Curl_sockerrno
Daniel Stenberg [Fri, 5 May 2006 10:24:27 +0000 (10:24 +0000)]
additional renames of Curl_ourerrno => Curl_sockerrno

18 years agoRoland Blom filed bug report #1481217
Daniel Stenberg [Thu, 4 May 2006 22:39:47 +0000 (22:39 +0000)]
Roland Blom filed bug report #1481217
(http://curl.haxx.se/bug/view.cgi?id=1481217), with follow-ups by Michele Bini
and David Byron. libcurl previously wrongly used GetLastError() on windows to
get error details after socket-related function calls, when it really should
use WSAGetLastError() instead.

When changing to this, the former function Curl_ourerrno() is now instead
called Curl_sockerrno() as it is necessary to only use it to get errno from
socket-related functions as otherwise it won't work as intended on Windows.

18 years agoMark Eichin submitted bug report #1480821
Daniel Stenberg [Thu, 4 May 2006 06:00:40 +0000 (06:00 +0000)]
Mark Eichin submitted bug report #1480821
(http://curl.haxx.se/bug/view.cgi?id=1480821) He found and identified a
problem with how libcurl dealt with GnuTLS and a case where gnutls returned
GNUTLS_E_AGAIN indicating it would block. It would then return an unexpected
return code, making Curl_ssl_send() confuse the upper layer - causing random
28 bytes trash data to get inserted in the transfered stream.

The proper fix was to make the Curl_gtls_send() function return the proper
return codes that the callers would expect. The Curl_ossl_send() function
already did this.

18 years agomoved the curl_off_t check to within the --enable-debug block where it belongs since...
Daniel Stenberg [Wed, 3 May 2006 22:39:49 +0000 (22:39 +0000)]
moved the curl_off_t check to within the --enable-debug block where it belongs since it is a somewhat ugly hack

18 years agoNick Mathewson added the ARES_OPT_SOCK_STATE_CB option that when set makes
Daniel Stenberg [Wed, 3 May 2006 06:11:44 +0000 (06:11 +0000)]
Nick Mathewson added the ARES_OPT_SOCK_STATE_CB option that when set makes
c-ares call a callback on socket state changes. A better way than the
ares_getsock() to get full control over the socket state.

18 years agocurl-config got a --checkfor option
Daniel Stenberg [Tue, 2 May 2006 22:48:22 +0000 (22:48 +0000)]
curl-config got a --checkfor option

18 years agoMake this code use the proper pointers
Daniel Stenberg [Tue, 2 May 2006 09:19:31 +0000 (09:19 +0000)]
Make this code use the proper pointers