5 \___|\___/|_| \_\_____|
9 Daniel Stenberg (23 Mar 2010)
10 - Hauke Duden provided an example program that made the multi interface crash.
11 His example simply used the multi interface and did first one FTP transfer
12 and after completion it used a second easy handle and did another FTP
13 transfer on the same FTP server.
15 This triggered a bug in the "delayed easy handle kill" system that curl
16 uses: when an FTP connection is left alive it must keep an easy handle
17 around internally - only for the purpose of having an easy handle when it
18 later disconnects it. The code assumed that when the easy handle was removed
19 and an internal reference was made, that version could be killed later on
20 when a new easy handle came using the same connection. This was wrong as
21 Hauke's example showed that the removed handle wasn't killed for real until
22 later. This caused a double close attempt => segfault.
24 Daniel Stenberg (22 Mar 2010)
25 - Thomas Lopatic fixed the alarm()-based DNS timeout:
27 Looking at the code of Curl_resolv_timeout() in hostip.c, I think that in
28 case of a timeout, the signal handler for SIGALRM never gets removed. I
29 think that in my case it gets executed at some point later on when execution
30 has long left Curl_resolv_timeout() or even the cURL library.
32 The code that is jumped to with siglongjmp() simply sets the error message
33 to "name lookup timed out" and then returns with CURLRESOLV_ERROR. I guess
34 that instead of simply returning without cleaning up, the code should have a
35 goto that jumps to the spot right after the call to Curl_resolv().
37 Kamil Dudka (22 Mar 2010)
38 - Douglas Steinwand contributed a patch fixing insufficient initialization in
39 Curl_clone_ssl_config()
41 Daniel Stenberg (21 Mar 2010)
42 - Ben Greear improved TFTP: the error code returning and the treatment
43 of TSIZE == 0 when uploading.
45 - We've switched from CVS to git. See http://curl.haxx.se/source.html
47 Kamil Dudka (19 Mar 2010)
48 - Improved Curl_read() to not ignore the error returned from Curl_ssl_recv().
50 Daniel Stenberg (15 Mar 2010)
51 - Constantine Sapuntzakis brought a patch:
53 The problem mentioned on Dec 10 2009
54 (http://curl.haxx.se/bug/view.cgi?id=2905220) was only partially fixed.
55 Partially because an easy handle can be associated with many connections in
56 the cache (e.g. if there is a redirect during the lifetime of the easy
57 handle). The previous patch only cleaned up the first one. The new fix now
58 removes the easy handle from all connections, not just the first one.
60 Daniel Stenberg (6 Mar 2010)
61 - Ben Greear brought a patch that fixed the rate limiting logic for TFTP when
62 the easy interface was used.
64 Daniel Stenberg (5 Mar 2010)
65 - Daniel Johnson provided fixes for building curl with the clang compiler.
68 - Constantine Sapuntzakis detected and fixed a double free in builds done
69 with threaded resolver enabled (Windows default configuration) that would
70 get triggered when a curl handle is closed while doing DNS resolution.
72 Daniel Stenberg (2 Mar 2010)
73 - [Daniel Johnson] I've been trying to build libcurl with clang on Darwin and
74 ran into some issues with the GSSAPI tests in configure.ac. The tests first
75 try to determine the include dirs and libs and set CPPFLAGS and LIBS
76 accordingly. It then checks for the headers and finally sets LIBS a second
77 time, causing the libs to be included twice. The first setting of LIBS seems
78 redundant and should be left out, since the first part is otherwise just
79 about finding headers.
81 My second issue is that 'krb5-config --libs gssapi' on Darwin is less than
82 useless and returns junk that, while it happens to work with gcc, causes
83 clang to choke. For example, --libs returns $CFLAGS along with the libs,
84 which is really retarded. Simply setting 'LIBS="$LIBS -lgssapi_krb5
85 -lresolv"' on Darwin is sufficient.
87 - Based on patch provided by Jacob Moshenko, the transfer logic now properly
88 makes sure that when using sub-second timeouts, there's no final bad 1000ms
89 wait. Previously, a sub-second timeout would often make the elapsed time end
90 up the time rounded up to the nearest second (e.g. 1s for 200ms timeout)
92 - Andrei Benea filed bug report #2956698 and pointed out that the
93 CURLOPT_CERTINFO feature leaked memory due to a missing OpenSSL function
94 call. He provided the patch to fix it too.
96 http://curl.haxx.se/bug/view.cgi?id=2956698
98 - Markus Duft pointed out in bug #2961796 that even though Interix has a
99 poll() function it doesn't quite work the way we want it so we must disable
100 it, and he also provided a patch for it.
102 http://curl.haxx.se/bug/view.cgi?id=2961796
104 - Made the pingpong timeout code properly deal with the response timeout AND
105 the global timeout if set. Also, as was reported in the bug report #2956437
106 by Ryan Chan, the time stamp to use as basis for the per command timeout was
107 not set properly in the DONE phase for FTP (and not for SMTP) so I fixed
108 that just now. This was a regression compared to 7.19.7 due to the
109 conversion of FTP code over to the generic pingpong concepts.
111 http://curl.haxx.se/bug/view.cgi?id=2956437
113 Daniel Stenberg (1 Mar 2010)
114 - Ben Greear provided an update for TFTP that fixes upload.
116 - Wesley Miaw reported bug #2958179 which identified a case of looping during
117 OpenSSL based SSL handshaking even though the multi interface was used and
118 there was no good reason for it.
120 http://curl.haxx.se/bug/view.cgi?id=2958179
122 Daniel Stenberg (26 Feb 2010)
123 - Pat Ray in bug #2958474 pointed out an off-by-one case when receiving a
124 chunked-encoding trailer.
126 http://curl.haxx.se/bug/view.cgi?id=2958474
128 Daniel Fandrich (25 Feb 2010)
129 - Fixed a couple of out of memory leaks and a segfault in the SMTP & IMAP code.
131 Yang Tse (25 Feb 2010)
132 - I fixed bug report #2958074 indicating
133 (http://curl.haxx.se/bug/view.cgi?id=2958074) that curl on Windows with
134 option --trace-time did not use local time when timestamping trace lines.
135 This could also happen on other systems depending on time souurce.
137 Patrick Monnerat (22 Feb 2010)
138 - Proper handling of STARTTLS on SMTP, taking CURLUSESSL_TRY into account.
139 - SMTP falls back to RFC821 HELO when EHLO fails (and SSL is not required).
140 - Use of true local host name (i.e.: via gethostname()) when available, as
141 default argument to SMTP HELO/EHLO.
142 - Test case 804 for HELO fallback.
144 Daniel Stenberg (20 Feb 2010)
145 - Fixed the SMTP compliance by making sure RCPT TO addresses are specified
146 properly in angle brackets. Recipients provided with CURLOPT_MAIL_RCPT now
147 get angle bracket wrapping automatically by libcurl unless the recipient
148 starts with an angle bracket as then the app is assumed to deal with that
151 - I made the SMTP code expect a 250 response back from the server after the
152 full DATA has been sent, and I modified the test SMTP server to also send
153 that response. As usual, the DONE operation that is made after a completed
154 transfer is still not doable in a non-blocking way so this waiting for 250
155 is unfortunately made blockingly.
157 Yang Tse (14 Feb 2010)
158 - Overhauled test suite getpart() function. Fixing potential out of bounds
159 stack and memory overwrites triggered with huge test case definitions.
161 Daniel Stenberg (13 Feb 2010)
162 - Martin Hager reported and fixed a problem with a missing quote in libcurl.m4
164 (http://curl.haxx.se/bug/view.cgi?id=2951319)
166 - Tom Donovan fixed the CURL_FORMAT_* defines when building with cmake.
168 (http://curl.haxx.se/bug/view.cgi?id=2951269)
170 Daniel Stenberg (12 Feb 2010)
171 - Jack Zhang reported a problem with SMTP: we wrongly used multiple addresses
172 in the same RCPT TO line, when they should be sent in separate single
173 commands. I updated test case 802 to verify this.
175 - I also fixed a bad use of my_setopt_str() of CURLOPT_MAIL_RCPT in the curl
176 tool which made it try to output it as string for the --libcurl feature
177 which could lead to crashes.
179 Yang Tse (11 Feb 2010)
180 - Steven M. Schweda fixed VMS builder bad behavior when used in a batch job,
181 removed obsolete batch_compile.com and defines.com and updated VMS readme.
183 Version 7.20.0 (9 February 2010)
185 Daniel Stenberg (9 Feb 2010)
186 - When downloading compressed content over HTTP and the app asked libcurl to
187 automatically uncompress it with the CURLOPT_ENCODING option, libcurl could
188 wrongly provide the callback with more data than the maximum documented
189 amount. An application could thus get tricked into badness if the maximum
190 limit was trusted to be enforced by libcurl itself (as it is documented).
192 This is further detailed and explained in the libcurl security advisory
195 http://curl.haxx.se/docs/adv_20100209.html
197 Daniel Fandrich (3 Feb 2010)
198 - Changed the Watcom makefiles to make them easier to keep in sync with
199 Makefile.inc since that can't be included directly.
201 Yang Tse (2 Feb 2010)
202 - Symbol CURL_FORMAT_OFF_T now obsoleted, will be removed in a future release,
203 symbol will not be available when building with CURL_NO_OLDIES defined. Use
204 of CURL_FORMAT_CURL_OFF_T is preferred since 7.19.0
206 Daniel Stenberg (1 Feb 2010)
207 - Using the multi_socket API, it turns out at times it seemed to "forget"
208 connections (which caused a hang). It turned out to be an existing (7.19.7)
209 bug in libcurl (that's been around for a long time) and it happened like
212 The app calls curl_multi_add_handle() to add a new easy handle, libcurl will
213 then set it to timeout in 1 millisecond so libcurl will tell the app about
216 The app's timeout fires off that there's a timeout, the app calls libcurl as
217 we so often document it:
220 res = curl_multi_socket_action(... TIMEOUT ...);
221 } while(CURLM_CALL_MULTI_PERFORM == res);
223 And this is the problem number one:
225 When curl_multi_socket_action() is called with no specific handle, but only
226 a timeout-action, it will *only* perform actions within libcurl that are
227 marked to run at this time. In this case, the request would go from INIT to
228 CONNECT and return CURLM_CALL_MULTI_PERFORM. When the app then calls libcurl
229 again, there's no timer set for this handle so it remains in the CONNECT
230 state. The CONNECT state is a transitional state in libcurl so it reports no
231 sockets there, and thus libcurl never tells the app anything more about that
232 easy handle/connection.
234 libcurl _does_ set a 1ms timeout for the handle at the end of
235 multi_runsingle() if it returns CURLM_CALL_MULTI_PERFORM, but since the loop
236 is instant the new job is not ready to run at that point (and there's no
237 code that makes libcurl call the app to update the timout for this new
238 timeout). It will simply rely on that some other timeout will trigger later
239 on or that something else will update the timeout callback. This makes the
240 bug fairly hard to repeat.
242 The fix made to adress this issue:
244 We introduce a loop in lib/multi.c around all calls to multi_runsingle() and
245 simply check for CURLM_CALL_MULTI_PERFORM internally. This has the added
246 benefit that this goes in line with my long-term wishes to get rid of the
247 CURLM_CALL_MULTI_PERFORM all together from the public API.
249 The downside of this fix, is that the counter we return in 'running_handles'
250 in several of our public functions then gets a slightly new and possibly
251 confusing behavior during times:
253 If an app adds a handle that fails to connect (very quickly) it may just
254 as well never appear as a 'running_handle' with this fix. Previously it
255 would first bump the counter only to get it decreased again at next call.
256 Even I have used that change in handle counter to signal "end of a
257 transfer". The only *good* way to find the end of a individual transfer
258 is calling curl_multi_info_read() to see if it returns one.
260 Of course, if the app previously did the looping before it checked the
261 counter, it really shouldn't be any new effect.
263 Yang Tse (26 Jan 2010)
264 - Constantine Sapuntzakis' and Joshua Kwan's work done in the last four months
265 relative to the asynchronous DNS lookups, along with with some integration
266 adjustments I have done are finally committed to CVS.
268 Currently these enhancements will benefit builds done using c-ares on any
269 platform as well as Windows builds using the default threaded resolver.
271 This release does not make generally available POSIX threaded DNS lookups
272 yet. There is no configure option to enable this feature yet. It is possible
273 to experimantally try this feature running configure with compiler flags that
274 make simultaneous definition of preprocessor symbols USE_THREADS_POSIX and
275 HAVE_PTHREAD_H, as well as whatever reentrancy compiler flags and linker ones
276 are required to link and properly use pthread_* functions on each platform.
278 Daniel Stenberg (26 Jan 2010)
279 - Mike Crowe made libcurl return CURLE_COULDNT_RESOLVE_PROXY when it is the
280 proxy that cannot be resolved when using c-ares. This matches the behaviour
281 when not using c-ares.
283 Björn Stenberg (23 Jan 2010)
284 - Added a new flag: -J/--remote-header-name. This option tells the
285 -O/--remote-name option to use the server-specified Content-Disposition
286 filename instead of extracting a filename from the URL.
288 Daniel Stenberg (21 Jan 2010)
289 - Chris Conroy brought support for RTSP transfers, and with it comes 8(!) new
290 libcurl options for controlling what to get and how to receive posssibly
291 interleaved RTP data.
293 Daniel Stenberg (20 Jan 2010)
294 - As was pointed out on the http-state mailing list, the order of cookies in a
295 HTTP Cookie: header _needs_ to be sorted on the path length in the cases
296 where two cookies using the same name are set more than once using
297 (overlapping) paths. Realizing this, identically named cookies must be
298 sorted correctly. But detecting only identically named cookies and take care
299 of them individually is harder than just to blindly and unconditionally sort
300 all cookies based on their path lengths. All major browsers also already do
301 this, so this makes our behavior one step closer to them in the cookie area.
303 Test case 8 was the only one that broke due to this change and I updated it
306 Daniel Stenberg (19 Jan 2010)
307 - David McCreedy brought a fix and a new test case (129) to make libcurl work
308 again when downloading files over FTP using ASCII and it turns out that the
309 final size of the file is not the same as the initial size the server
310 reported. This is very common since servers don't take the newline
311 conversions into account.
313 Kamil Dudka (14 Jan 2010)
314 - Suppressed side effect of OpenSSL configure checks, which prevented NSS from
315 being properly detected under certain circumstances. It had been caused by
316 strange behavior of pkg-config when handling PKG_CONFIG_LIBDIR. pkg-config
317 distinguishes among empty and non-existent environment variable in that case.
319 Daniel Stenberg (12 Jan 2010)
320 - Gil Weber reported a peculiar flaw with the multi interface when doing SFTP
321 transfers: curl_multi_fdset() would return -1 and not set and file
322 descriptors several times during a transfer of a single file. It turned out
323 to be due to two different flaws now fixed. Gil's excellent recipe helped me
326 Daniel Stenberg (11 Jan 2010)
327 - Made sure that the progress callback is repeatedly called at a regular
328 interval even during very slow connects.
330 - The tests/runtests.pl script now checks to see if the test case that runs is
331 present in the tests/data/Makefile.am and outputs a notice message on the
332 screen if not. Each test file has to be included in that Makefile.am to get
333 included in release archives and forgetting to add files there is a common
334 mistake. This is an attempt to make it harder to forget.
336 Daniel Stenberg (9 Jan 2010)
337 - Johan van Selst found and fixed a OpenSSL session ref count leak:
339 ossl_connect_step3() increments an SSL session handle reference counter on
340 each call. When sessions are re-used this reference counter may be
341 incremented many times, but it will be decremented only once when done (by
342 Curl_ossl_session_free()); and the internal OpenSSL data will not be freed
343 if this reference count remains positive. When a session is re-used the
344 reference counter should be corrected by explicitly calling
345 SSL_SESSION_free() after each consecutive SSL_get1_session() to avoid
346 introducing a memory leak.
348 (http://curl.haxx.se/bug/view.cgi?id=2926284)
350 Daniel Stenberg (7 Jan 2010)
351 - Make sure the progress callback is called repeatedly even during very slow
352 name resolves when c-ares is used for resolving.
354 Claes Jakobsson (6 Jan 2010)
355 - Julien Chaffraix fixed so that the fragment part in an URL is not sent
356 to the server anymore.
358 Kamil Dudka (3 Jan 2010)
359 - Julien Chaffraix eliminated a duplicated initialization in singlesocket().
361 Daniel Stenberg (2 Jan 2010)
362 - Make curl support --ssl and --ssl-reqd instead of the previous FTP-specific
363 versions --ftp-ssl and --ftp-ssl-reqd as these options are now used to
364 control SSL/TLS for IMAP, POP3 and SMTP as well in addition to FTP. The old
365 option names are still working but the new ones are the ones listed and
368 Daniel Stenberg (1 Jan 2010)
369 - Ingmar Runge enhanced libcurl's FTP engine to support the PRET command. This
370 command is a special "hack" used by the drftpd server, but even though it is
371 a custom extension I've deemed it fine to add to libcurl since this server
372 seems to survive and people keep using it and want libcurl to support
373 it. The new libcurl option is named CURLOPT_FTP_USE_PRET, and it is also
374 usable from the curl tool with --ftp-pret. Using this option on a server
375 that doesn't support this command will make libcurl fail.
377 I added test cases 1107 and 1108 to verify the functionality.
379 The PRET command is documented at
380 http://www.drftpd.org/index.php/Distributed_PASV
382 Yang Tse (30 Dec 2009)
383 - Steven M. Schweda improved VMS build system, and Craig A. Berry helped
384 with the patch and testing.
386 Daniel Stenberg (26 Dec 2009)
387 - Renato Botelho and Peter Pentchev brought a patch that makes the libcurl
388 headers work correctly even on FreeBSD systems before v8.
390 (http://curl.haxx.se/bug/view.cgi?id=2916915)
392 Daniel Stenberg (17 Dec 2009)
393 - David Byron fixed Curl_ossl_cleanup to actually call ENGINE_cleanup when
396 - Follow-up fix for the proxy fix I did for Jon Nelson's bug. It turned out I
397 was a bit too quick and broke test case 1101 with that change. The order of
398 some of the setups is sensitive. I now changed it slightly again to make
399 sure we do them in this order:
401 1 - parse URL and figure out what protocol is used in the URL
402 2 - prepend protocol:// to URL if missing
403 3 - parse name+password off URL, which needs to know what protocol is used
404 (since only some allows for name+password in the URL)
405 4 - figure out if a proxy should be used set by an option
406 5 - if no proxy option, check proxy environment variables
407 6 - run the protocol-specific setup function, which needs to have the proxy
410 Daniel Stenberg (15 Dec 2009)
411 - Jon Nelson found a regression that turned out to be a flaw in how libcurl
412 detects and uses proxies based on the environment variables. If the proxy
413 was given as an explicit option it worked, but due to the setup order
414 mistake proxies would not be used fine for a few protocols when picked up
415 from '[protocol]_proxy'. Obviously this broke after 7.19.4. I now also added
416 test case 1106 that verifies this functionality.
418 (http://curl.haxx.se/bug/view.cgi?id=2913886)
420 Daniel Stenberg (12 Dec 2009)
421 - IMAP, POP3 and SMTP support and their TLS versions (including IMAPS, POP3S
422 and SMTPS) are now supported. The current state may not yet be solid, but
423 the foundation is in place and the test suite has some initial support for
424 these protocols. Work will now persue to make them nice libcurl citizens
427 The work with supporting these new protocols was sponsored by
428 networking4all.com - thanks!
430 Daniel Stenberg (10 Dec 2009)
431 - Siegfried Gyuricsko found out that the curl manual said --retry would retry
432 on FTP errors in the transient 5xx range. Transient FTP errors are in the
433 4xx range. The code itself only tried on 5xx errors that occured _at login_.
434 Now the retry code retries on all FTP transfer failures that ended with a
437 (http://curl.haxx.se/bug/view.cgi?id=2911279)
439 - Constantine Sapuntzakis figured out a case which would lead to libcurl
440 accessing alredy freed memory and thus crash when using HTTPS (with
441 OpenSSL), multi interface and the CURLOPT_DEBUGFUNCTION and a certain order
442 of cleaning things up. I fixed it.
444 (http://curl.haxx.se/bug/view.cgi?id=2905220)
446 Daniel Stenberg (7 Dec 2009)
447 - Martin Storsjo made libcurl use the Expect: 100-continue header for posts
448 with unknown size. Previously it was only used for posts with a known size
449 larger than 1024 bytes.
451 Daniel Stenberg (1 Dec 2009)
452 - If the Expect: 100-continue header has been set by the application through
453 curl_easy_setopt with CURLOPT_HTTPHEADER, the library should set
454 data->state.expect100header accordingly - the current code (in 7.19.7 at
455 least) doesn't handle this properly. Martin Storsjo provided the fix!
457 Yang Tse (28 Nov 2009)
458 - Added Diffie-Hellman parameters to several test harness certificate files in
459 PEM format. Required by several stunnel versions used by our test harness.
461 Daniel Stenberg (28 Nov 2009)
462 - Markus Koetter provided a polished and updated version of Chad Monroe's TFTP
463 rework patch that now integrates TFTP properly into libcurl so that it can
464 be used non-blocking with the multi interface and more. BLKSIZE also works.
466 The --tftp-blksize option was added to allow setting the TFTP BLKSIZE from
469 Daniel Stenberg (26 Nov 2009)
470 - Extended and fixed the change I did on Dec 11 for the the progress
471 meter/callback during FTP command/response sequences. It turned out it was
472 really lame before and now the progress meter SHOULD get called at least
475 Daniel Stenberg (23 Nov 2009)
476 - Bjorn Augustsson reported a bug which made curl not report any problems even
477 though it failed to write a very small download to disk (done in a single
478 fwrite call). It turned out to be because fwrite() returned success, but
479 there was insufficient error-checking for the fclose() call which tricked
480 curl to believe things were fine.
482 Yang Tse (23 Nov 2009)
483 - David Byron modified Makefile.dist vc8 and vc9 targets in order to allow
484 finer granularity control when generating src and lib makefiles.
486 Yang Tse (22 Nov 2009)
487 - I modified configure to force removal of the curlbuild.h file included in
488 distribution tarballs for use by non-configure systems. As intended, this
489 would get overwriten when doing in-tree builds. But VPATH builds would end
490 having two curlbuild.h files, one in the source tree and another in the
491 build tree. With the modification I introduced 5 Nov 2009 this could become
492 an issue when running libcurl's test suite.
494 Daniel Stenberg (20 Nov 2009)
495 - Constantine Sapuntzakis identified a write after close, as the sockets were
496 closed by libcurl before the SSL lib were shutdown and they may write to its
497 socket. Detected to at least happen with OpenSSL builds.
499 - Jad Chamcham pointed out a bug with connection re-use. If a connection had
500 CURLOPT_HTTPPROXYTUNNEL enabled over a proxy, a subsequent request using the
501 same proxy with the tunnel option disabled would still wrongly re-use that
502 previous connection and the outcome would only be badness.
504 Yang Tse (18 Nov 2009)
505 - I modified the memory tracking system to make it intolerant with zero sized
506 malloc(), calloc() and realloc() function calls.
508 Daniel Stenberg (17 Nov 2009)
509 - Constantine Sapuntzakis provided another fix for the DNS cache that could
510 end up with entries that wouldn't time-out:
512 1. Set up a first web server that redirects (307) to a http://server:port
514 2. Have curl connect to the first web server using curl multi
516 After the curl_easy_cleanup call, there will be curl dns entries hanging
517 around with in_use != 0.
519 (http://curl.haxx.se/bug/view.cgi?id=2891591)
521 - Marc Kleine-Budde fixed: curl saved the LDFLAGS set during configure into
522 its pkg-config file. So -Wl stuff ended up in the .pc file, which is really
523 bad, and breaks if there are multiple -Wl in our LDFLAGS (which are in
524 PTXdist). bug #2893592 (http://curl.haxx.se/bug/view.cgi?id=2893592)
526 Kamil Dudka (15 Nov 2009)
527 - David Byron improved the configure script to use pkg-config to find OpenSSL
528 (and in particular the list of required libraries) even if a path is given
529 as argument to --with-ssl
531 Yang Tse (15 Nov 2009)
532 - I removed enable-thread / disable-thread configure option. These were only
533 placebo options. The library is always built as thread safe as possible on
536 Claes Jakobsson (14 Nov 2009)
537 - curl-config now accepts '--configure' to see what arguments was
538 passed to the configure script when building curl.
540 Daniel Stenberg (14 Nov 2009)
541 - Claes Jakobsson restored the configure functionality to detect NSS when
542 --with-nss is set but not "yes".
544 I think we can still improve that to check for pkg-config in that path etc,
545 but at least this patch brings back the same functionality we had before.
547 - Camille Moncelier added support for the file type SSL_FILETYPE_ENGINE for
548 the client certificate. It also disable the key name test as some engines
549 can select a private key/cert automatically (When there is only one key
550 and/or certificate on the hardware device used by the engine)
552 Yang Tse (14 Nov 2009)
553 - Constantine Sapuntzakis provided the fix that ensures that an SSL connection
554 won't be reused unless protection level for peer and host verification match.
556 I refactored how preprocessor symbol _THREAD_SAFE definition is done.
558 Kamil Dudka (12 Nov 2009)
559 - Kevin Baughman provided a fix preventing libcurl-NSS from crash on doubly
560 closed NSPR descriptor. The issue was hard to find, reported several times
561 before and always closed unresolved. More info at the RH bug:
562 https://bugzilla.redhat.com/534176
564 - libcurl-NSS now tries to reconnect with TLS disabled in case it detects
565 a broken TLS server. However it does not happen if SSL version is selected
566 manually. The approach was originally taken from PSM. Kaspar Brand helped me
567 to complete the patch. Original bug reports:
568 https://bugzilla.redhat.com/525496
569 https://bugzilla.redhat.com/527771
571 Yang Tse (12 Nov 2009)
572 - I modified configure script to make the getaddrinfo function check also
573 verify if the function is thread safe.
575 Yang Tse (11 Nov 2009)
576 - Marco Maggi reported that compilation failed when configured --with-gssapi
577 and GNU GSS installed due to a missing mutual exclusion of header files in
578 the Kerberos 5 code path. He also verified that my patch worked for him.
580 Daniel Stenberg (11 Nov 2009)
581 - Constantine Sapuntzakis posted bug #2891595
582 (http://curl.haxx.se/bug/view.cgi?id=2891595) which identified how an entry
583 in the DNS cache would linger too long if the request that added it was in
584 use that long. He also provided the patch that now makes libcurl capable of
585 still doing a request while the DNS hash entry may get timed out.
587 - Christian Schmitz noticed that the progress meter/callback was not properly
588 used during the FTP connection phase (after the actual TCP connect), while
589 it of course should be. I also made the speed check get called correctly so
590 that really slow servers will trigger that properly too.
592 Kamil Dudka (5 Nov 2009)
593 - Dropped misleading timeouts in libcurl-NSS and made sure the SSL socket works
594 in non-blocking mode.
596 Yang Tse (5 Nov 2009)
597 - I removed leading 'curl' path on the 'curlbuild.h' include statement in
598 curl.h, adjusting auto-makefiles include path, to enhance portability to
599 OS's without an orthogonal directory tree structure such as OS/400.
601 Daniel Stenberg (4 Nov 2009)
602 - I fixed several problems with the transfer progress meter. It showed the
603 wrong percentage for small files, most notable for <1000 bytes and could
604 easily end up showing more than 100% at the end. It also didn't show any
605 percentage, transfer size or estimated transfer times when transferring
608 Version 7.19.7 (4 November 2009)
610 Daniel Stenberg (2 Nov 2009)
611 - As reported independent by both Stan van de Burgt and Didier Brisebourg,
612 CURLINFO_SIZE_DOWNLOAD (the -w variable size_download) didn't work when
613 getting data from ldap!
615 Daniel Stenberg (31 Oct 2009)
616 - Gabriel Kuri reported a problem with CURLINFO_CONTENT_LENGTH_DOWNLOAD if the
617 download was 0 bytes, as libcurl would then return the size as unknown (-1)
618 and not 0. I wrote a fix and test case 566 to verify it.
620 Daniel Stenberg (30 Oct 2009)
621 - Liza Alenchery mentioned a problem with re-used SCP connection when a bad
622 auth is used, as it caused a crash. I failed to repeat the issue, but still
623 made a change that now forces the TCP connection used for a freed SCP
624 session to get closed and not be re-used.
626 - "Tom" posted a bug report that mentioned how libcurl did wrong when doing a
627 POST using a read callback, with Digest authentication and
628 "Transfer-Encoding: chunked" enforced. I would then cause the first request
629 to be wrongly sent and then basically hang until the server closed the
630 connection. I fixed the problem and added test case 565 to verify it.
632 Daniel Stenberg (25 Oct 2009)
633 - Dima Barsky made the curl cookie parser accept cookies even with blank or
634 unparsable expiry dates and then treat them as session cookies - previously
635 libcurl would reject cookies with a date format it couldn't parse. Research
636 shows that the major browser treat such cookies as session cookies. I
637 modified test 8 and 31 to verify this.
639 Daniel Stenberg (21 Oct 2009)
640 - Attempt to use pkg-config for finding out libssh2 installation details
643 - A patch in bug report #2883177 (http://curl.haxx.se/bug/view.cgi?id=2883177)
644 by Johan van Selst introduced the --crlfile option to curl, which makes curl
645 tell libcurl about a file with CRL (certificate revocation list) data to
648 Daniel Stenberg (18 Oct 2009)
649 - Ray Dassen provided a patch in Debian's bug tracker (bug number #551461)
650 that now makes curl_getdate(3) actually handles RFC 822 formatted dates that
651 use the "single letter military timezones".
652 http://www.rfc-ref.org/RFC-TEXTS/822/chapter5.html has the details.
654 - Fixed memory leak in the SCP/SFTP code as it never freed the knownhosts
657 - John Dennis filed bug report #2873666
658 (http://curl.haxx.se/bug/view.cgi?id=2873666) which identified a problem
659 which made libcurl loop infinitely when given incorrect credentials when
660 using HTTP GSS negotiate authentication. He also provided a small and simple
663 - Kevin Baughman found a double close() problem with libcurl-NSS, as when
664 libcurl called NSS to close the SSL "session" it also closed the actual
667 Yang Tse (17 Oct 2009)
668 - Bug report #2866724 indicated
669 (http://curl.haxx.se/bug/view.cgi?id=2866724) that curl on Windows failed
670 when writing files whose file names originally contained characters which
671 are not valid for file names on Windows. Dan Fandrich provided an initial
672 patch and another revised one to fix this issue.
674 Daniel Stenberg (1 Oct 2009)
675 - Tom Mueller correctly reported in bug report #2870221
676 (http://curl.haxx.se/bug/view.cgi?id=2870221) that libcurl returned an
677 incorrect return code from the internal trynextip() function which caused
678 him grief. This is a regression that was introduced in 7.19.1 and I find it
679 strange it hasn't hit us harder, but I won't persue into figuring out
682 - Constantine Sapuntzakis: The current implementation will always set
683 SO_SNDBUF to CURL_WRITE_SIZE even if the SO_SNDBUF starts out larger. The
684 patch doesn't do a setsockopt if SO_SNDBUF is already greater than
685 CURL_WRITE_SIZE. This should help folks who have set up their computer with
688 Daniel Stenberg (27 Sep 2009)
689 - I introduced a maximum limit for received HTTP headers. It is controlled by
690 the define CURL_MAX_HTTP_HEADER which is even exposed in the public header
691 file to allow for users to fairly easy rebuild libcurl with a modified
692 limit. The rationale for a fixed limit is that libcurl is realloc()ing a
693 buffer to be able to put a full header into it, so that it can call the
694 header callback with the entire header, but that also risk getting it into
695 trouble if a server by mistake or willingly sends a header that is more or
696 less without an end. The limit is set to 100K.
698 Daniel Stenberg (26 Sep 2009)
699 - John P. McCaskey posted a bug report that showed how libcurl did wrong when
700 saving received cookies with no given path, if the path in the request had a
701 query part. That is means a question mark (?) and characters on the right
702 side of that. I wrote test case 1105 and fixed this problem.
704 Kamil Dudka (26 Sep 2009)
705 - Implemented a protocol independent way to specify blocking direction, used by
706 transfer.c for blocking. It is currently used only by SCP and SFTP protocols.
707 This enhancement resolves an issue with 100% CPU usage during SFTP upload,
710 Daniel Stenberg (25 Sep 2009)
711 - Chris Mumford filed bug report #2861587
712 (http://curl.haxx.se/bug/view.cgi?id=2861587) identifying that libcurl used
713 the OpenSSL function X509_load_crl_file() wrongly and failed if it would
714 load a CRL file with more than one certificate within. This is now fixed.
716 Daniel Stenberg (16 Sep 2009)
717 - Sven Anders reported that we introduced a cert verfication flaw for OpenSSL-
718 powered libcurl in 7.19.6. If there was a X509v3 Subject Alternative Name
719 field in the certficate it had to match and so even if non-DNS and non-IP
720 entry was present it caused the verification to fail.
722 Daniel Fandrich (15 Sep 2009)
723 - Moved the libssh2 checks after the SSL library checks. This helps when
724 statically linking since libssh2 needs the SSL library link flags to be
725 set up already to satisfy its dependencies. This wouldn't be necessary if
726 the libssh2 configure check was changed to use pkg-config since the
727 --static flag would add the dependencies automatically.
729 Yang Tse (14 Sep 2009)
730 - Revert Joshua Kwan's patch committed 11 Sep 2009.
732 Some systems poll function sets POLLHUP in revents without setting
733 POLLIN, and sets POLLERR without setting POLLIN and POLLOUT. In some
734 libcurl code execution paths this could trigger busy wait loops with
735 high CPU usage until a timeout condition aborted the loop.
737 The reverted patch addressed the above issue for a very specific case,
738 when awaiting c-ares to resolve. A libcurl-wide fix for Curl_poll now
741 Guenter Knauf (11 Sep 2009)
742 - Joshua Kwan provided a patch to pass POLLERR / POLLHUP back to c-ares.
743 This fixes a loop problem with high CPU usage.
745 Daniel Stenberg (10 Sep 2009)
746 - Claes Jakobsson fixed a problem with cookie expiry dates at exctly the epoch
747 start second "Thu Jan 1 00:00:00 GMT 1970" as the date parser then returns 0
748 which internally then is treated as a session cookie. That particular date
749 is now made to get the value of 1.
751 Daniel Stenberg (2 Sep 2009)
752 - Daniel Johnson found a flaw in the code converting sftp-errors to libcurl
755 Daniel Stenberg (1 Sep 2009)
756 - Peter Sylvester made a debug feature for Curl_resolv() that now will force
757 libcurl to resolve 'localhost' whatever name you use in the URL *if* you set
758 the --interface option to (exactly) "LocalHost". This will enable us to
759 write tests for custom hosts names but still use a local host server.
761 - configure now tries to use pkg-config for a number of sub-dependencies even
762 when cross-compiling. The key to success is then you properly setup
763 PKG_CONFIG_PATH before invoking configure.
765 I also improved how NSS is detected by trying nss-config if pkg-config isn't
766 present, and as a last resort just use the lib name and force the user to
767 setup the LIBS/LDFLAGS/CFLAGS etc properly. The previous last resort would
768 add a range of various libs that would almost never be quite correct.
770 Daniel Stenberg (31 Aug 2009)
771 - When using the multi interface with FTP and you asked for NOBODY, you did no
772 QUOTE commands and the request used the same path as the connection had
773 already changed to, it would decide that no commands would be necessary for
774 the "DO" action and that was not handled properly but libcurl would instead
777 Kamil Dudka (28 Aug 2009)
778 - Improved error message for not matching certificate subject name in
779 libcurl-NSS. Originally reported at:
780 https://bugzilla.redhat.com/show_bug.cgi?id=516056#c9
782 Patrick Monnerat (24 Aug 2009)
783 - Introduced a SYST-based test to properly set-up name format when dealing
784 with the OS/400 FTP server.
786 - Fixed an ftp_readresp() bug preventing detection of failing control socket
787 and causing FTP client to loop forever.
789 Daniel Stenberg (24 Aug 2009)
790 - Marc de Bruin pointed out that configure --with-gnutls=PATH didn't work
791 properly and provided a fix. http://curl.haxx.se/bug/view.cgi?id=2843008
793 - Eric Wong introduced support for the new option -T. (dot) that makes curl
794 read stdin in a non-blocking fashion. This also brings back -T- (minus) to
795 the previous blocking behavior since it could break stuff for people at
798 Michal Marek (21 Aug 2009)
799 - With CURLOPT_PROXY_TRANSFER_MODE, avoid sending invalid URLs like
800 ftp://example.com;type=i if the user specified ftp://example.com without the
803 Daniel Stenberg (21 Aug 2009)
804 - Andre Guibert de Bruet pointed out a missing return code check for a
805 strdup() that could lead to segfault if it returned NULL. I extended his
806 suggest patch to now have Curl_retry_request() return a regular return code
807 and better check that.
809 - Lots of good work by Krister Johansen, mostly related to pipelining:
811 Fix SIGSEGV on free'd easy_conn when pipe unexpectedly breaks
812 Fix data corruption issue with re-connected transfers
813 Fix use after free if we're completed but easy_conn not NULL
815 Kamil Dudka (13 Aug 2009)
816 - Changed NSS code to not ignore the value of ssl.verifyhost and produce more
817 verbose error messages. Originally reported at:
818 https://bugzilla.redhat.com/show_bug.cgi?id=516056
820 Daniel Stenberg (12 Aug 2009)
821 - Karl Moerder fixed the Makefile.vc* makefiles to include the new file
822 nonblock.c so that they work fine again
824 - I expanded test 517 with a bunch of more dates that originate from the
825 Chrome browser test suite. It turns out most of them get parsed the same
828 Version 7.19.6 (12 August 2009)
830 Daniel Stenberg (12 Aug 2009)
831 - Carsten Lange reported a bug and provided a patch for TFTP upload and the
832 sending of the TSIZE option. I don't like fixing bugs just hours before
833 a release, but since it was broken and the patch fixes this for him I decided
836 Daniel Stenberg (11 Aug 2009)
837 - Peter Sylvester made the HTTPS test server use specific certificates for
838 each test, so that the test suite can now be used to actually test the
839 verification of cert names etc. This made an error show up in the OpenSSL-
840 specific code where it would attempt to match the CN field even if a
841 subjectAltName exists that doesn't match. This is now fixed and verified
844 - Benbuck Nason posted the bug report #2835196
845 (http://curl.haxx.se/bug/view.cgi?id=2835196), fixing a few compiler
846 warnings when mixing ints and bools.
848 Daniel Fandrich (10 Aug 2009)
849 - Fixed a memory leak in the FTP code and an off-by-one heap buffer overflow.
851 Daniel Fandrich (9 Aug 2009)
852 - Fixed some memory leaks in the command-line tool that caused most of the
853 torture tests to fail.
855 Daniel Stenberg (2 Aug 2009)
856 - Curt Bogmine reported a problem with SNI enabled on a particular server. We
857 should introduce an option to disable SNI, but as we're in feature freeze
858 now I've addressed the obvious bug here (pointed out by Peter Sylvester): we
859 shouldn't try to enable SNI when SSLv2 or SSLv3 is explicitly selected.
860 Code for OpenSSL and GnuTLS was fixed. NSS doesn't seem to have a particular
861 option for SNI, or are we simply not using it?
863 Daniel Stenberg (1 Aug 2009)
864 - Scott Cantor posted the bug report #2829955
865 (http://curl.haxx.se/bug/view.cgi?id=2829955) mentioning the recent SSL cert
866 verification flaw found and exploited by Moxie Marlinspike. The presentation
867 he did at Black Hat is available here:
868 https://www.blackhat.com/html/bh-usa-09/bh-usa-09-archives.html#Marlinspike
870 Apparently at least one CA allowed a subjectAltName or CN that contain a
871 zero byte, and thus clients that assumed they would never have zero bytes
872 were exploited to OK a certificate that didn't actually match the site. Like
873 if the name in the cert was "example.com\0theatualsite.com", libcurl would
874 happily verify that cert for example.com.
876 libcurl now better uses the length of the extracted name, not using the zero
877 termination for getting the string length.
879 This fixing only made and needed in OpenSSL interfacing code.
881 - Tanguy Fautre pointed out that OpenSSL's function RAND_screen() (present
882 only in some OpenSSL installs - like on Windows) isn't thread-safe and we
883 agreed that moving it to the global_init() function is a decent way to deal
886 - Alexander Beedie provided the patch for a noproxy problem: If I have set
887 CURLOPT_NOPROXY to "*", or to a host that should not use a proxy, I actually
888 could still end up using a proxy if a proxy environment variable was set.
890 Daniel Stenberg (27 Jul 2009)
891 - All the quote options (CURLOPT_QUOTE, CURLOPT_POSTQUOTE and
892 CURLOPT_PREQUOTE) now accept a preceeding asterisk before the command to
893 send when using FTP, as a sign that libcurl shall simply ignore the response
894 from the server instead of treating it as an error. Not treating a 400+ FTP
895 response code as an error means that failed commands will not abort the
896 chain of commands, nor will they cause the connection to get disconnected.
898 Daniel Stenberg (26 Jul 2009)
899 - Johan van Selst posted bug report #2825989
900 (http://curl.haxx.se/bug/view.cgi?id=2825989) pointing out that
901 OpenSSL-powered libcurl didn't support the SHA-2 digest algorithm, and
902 provided the solution too: to use OpenSSL_add_all_algorithms() in addition
903 to the older SSLeay_* alternative. OpenSSL_add_all_algorithms was added in
906 Daniel Stenberg (23 Jul 2009)
907 - Added CURLOPT_SSH_KNOWNHOSTS, CURLOPT_SSH_KEYFUNCTION, CURLOPT_SSH_KEYDATA.
908 They introduce known_host support for SSH keys to libcurl. See docs for
909 details. Note that this feature depends on a new enough libssh2 version, to
910 be supported in libssh2 1.2 and later (or current git repo at this time).
912 Michal Marek (22 Jul 2009)
913 - David Binderman found a memory and fd leak in lib/gtls.c:load_file()
914 (https://bugzilla.novell.com/523919). When looking at the code, I found that
915 also the ptr pointer can leak.
917 Kamil Dudka (20 Jul 2009)
918 - Claes Jakobsson improved the support for client certificates handling in
919 NSS-powered libcurl. Now the client certificates can be selected
920 automatically by a NSS built-in hook. Additionally pre-login to all PKCS11
921 slots is no more performed. It used to cause problems with HW tokens.
923 - Fixed reference counting for NSS client certificates. Now the PEM reader
924 module should be always properly unloaded on Curl_nss_cleanup(). If the
925 unload fails though, libcurl will try to reuse the already loaded instance.
927 Daniel Fandrich (15 Jul 2009)
928 - Added nonblock.c to the non-automake makefiles (note that the dependencies
929 in the Watcom makefiles aren't quite correct).
931 Michal Marek (15 Jul 2009)
932 - Changed the description of CURLINFO_OS_ERRNO to make it clear that the
933 errno is not reset on success.
935 Guenter Knauf (14 Jul 2009)
936 - renamed generated config.h to curl_config.h to avoid any future clashes
937 with config.h from other projects.
939 Daniel Stenberg (9 Jul 2009)
940 - Eric Wong introduced curlx_nonblock() that the curl tool now (re-)uses for
941 setting a file descriptor non-blocking. Used by the functionality Eric
942 himself brough on June 15th.
944 Daniel Stenberg (8 Jul 2009)
945 - Constantine Sapuntzakis posted bug report #2813123
946 (http://curl.haxx.se/bug/view.cgi?id=2813123) and an a patch that fixes the
949 Url A is accessed using auth. Url A redirects to Url B (on a different
950 server0. Url B reuses a persistent connection. Url B has auth, even though
951 it's on a different server.
953 Note: if Url B does not reuse a persistent connection, auth is not sent.
957 data->state.first_host is not initialized becuase Curl_http_connect is not
958 called when a connection is reused.
962 move initialization of data->state.first_host to Curl_http. No code before
963 Curl_http uses data->state.first_host anyway.
965 Guenter Knauf (4 Jul 2009)
966 - Markus Koetter provided a patch to avoid getnameinfo() usage which broke a
967 couple of both IPv4 and IPv6 autobuilds.
969 Daniel Stenberg (29 Jun 2009)
970 - Markus Koetter made CURLOPT_FTPPORT (and curl's -P/--ftpport) support a port
971 range if given colon-separated after the host name/address part. Like
972 "192.168.0.1:2000-10000"
974 - Modified the separators used for CURLOPT_CERTINFO in multi-part outputs. I
975 don't know how they got wrong in the first place, but using this output
976 format makes it possible to quite easily separate the string into an array
979 Daniel Fandrich (16 June 2009)
980 - Added a few more compiler warning options for gcc.
982 Daniel Stenberg (16 Jun 2009)
983 - Reuven Wachtfogel made curl -o - properly produce a binary output on windows
984 (no newline translations). Use -B/--use-ascii if you rather get the ascii
987 Michal Marek (16 Jun 2009)
988 - When doing non-anonymous ftp via http proxies and the password is not
989 provided in the url, add it there (squid needs this).
991 Daniel Stenberg (15 Jun 2009)
994 This allows curl(1) to be used as a client-side tunnel for arbitrary stream
995 protocols by abusing chunked transfer encoding in both the HTTP request and
996 HTTP response. This requires server support for sending a response while a
997 request is still being read, of course.
999 If attempting to read from stdin returns EAGAIN, then we pause our sender.
1000 This leaves curl to attempt to read from the socket while reading from stdin
1001 (and thus sending) is paused.
1003 This change was needed to allow successfully tunneling the git protocol over
1004 HTTP (--no-buffer is needed, as well).
1006 Patrick Monnerat (15 Jun 2009)
1007 - Replaced use of standard C library rand()/srand() by our own pseudo-random
1010 Yang Tse (11 Jun 2009)
1011 - I adapted testcurl script to allow building test harness programs when
1012 cross-compiling for a *-*-mingw* host.
1014 Daniel Stenberg (10 Jun 2009)
1015 - Fabian Keil ran clang on the (lib)curl code, found a bunch of warnings and
1016 contributed a range of patches to fix them.
1018 Yang Tse (10 Jun 2009)
1019 - I introduced configure script option --enable-curldebug which now allows
1020 the decoupled enabling or disabling of the curl debug memory tracking
1021 feature from the --enable-debug option which no longer controls this.
1023 curl --version will list 'Debug' feature for debug enabled builds, and
1024 will list 'TrackMemory' feature for curl debug memory tracking capable
1025 builds. These features are independent and can be controlled when running
1026 the configure script. When --enable-debug is given both features will be
1027 enabled, unless some restriction prevents memory tracking from being used.
1029 Internally, definition of preprocessor symbol DEBUGBUILD restricts code
1030 which is only compiled for debug enabled builds. And symbol CURLDEBUG is
1031 used to differentiate code which is _only_ used for memory tracking.
1033 Yang Tse (9 Jun 2009)
1034 - Daniel Steinberg pointed out that Curl_FormInit() in formdata.c was not
1035 initializing the fread callback pointer and this triggered a compiler
1036 warning, also provided a friendly suggestion on how to fix it.
1038 Daniel Stenberg (8 Jun 2009)
1039 - Claes Jakobsson provided a patch for libcurl-NSS that fixed a bad refcount
1040 issue with client certs that caused issues like segfaults.
1041 http://curl.haxx.se/mail/lib-2009-05/0316.html
1043 - Triggered by bug report #2798852 and the patch in there, I fixed configure
1044 to detect gnutls build options with pkg-config only and not libgnutls-config
1045 anymore since GnuTLS has stopped distributing that tool. If an explicit path
1046 is given to configure, we will instead guess on how to link and use that
1047 lib. I did not use the patch from the bug report.
1049 Yang Tse (8 Jun 2009)
1050 - Igor Novoseltsev adjusted Makefile.vxworks to get sources and headers
1051 included from Makefile.inc, and provided docs\INSTALL VxWorks section.
1053 - I removed buildconf.bat from release and daily snapshot archives. This
1054 file is only for CVS tree checkout builds.
1056 Daniel Stenberg (8 Jun 2009)
1057 - Eric Wong fixed --no-buffer to actually switch off output buffering. Been
1060 Bill Hoffman (6 Jun 2009)
1061 - Added some cmake docs and fixed socklen_t in the build.
1063 Yang Tse (5 Jun 2009)
1064 - John E. Malmberg provided VMS specific patch: "This fixes an existing bug
1065 in urlglob.c where it was not converting the Curl Unix exit code to a VMS
1066 DCL compatible exit code. This fix required the enhancement described next.
1067 This also adds an enhancement to main.c so that when curl is run under a
1068 Unix shell like Bash on VMS, it will return the standard Unix exit codes
1069 and messages." And another patch for docs/examples.
1071 I introduced os-specific.c and os-specific.h for use in curl tool code
1072 and adjusted John E. Malmberg's patch placement to use these new files
1073 as an effort to prevent main.c from growing ad infinitum. Code already
1074 existing in main.c which is OS specific should be moved into these files.
1076 Daniel Stenberg (4 June 2009)
1077 - Setting the Content-Length: header from your app when you do a POST or PUT
1078 is almost always a VERY BAD IDEA. Yet there are still apps out there doing
1079 this, and now recently it triggered a bug/side-effect in libcurl as when
1080 libcurl sends a POST or PUT with NTLM, it sends an empty post first when it
1081 knows it will just get a 401/407 back. If the app then replaced the
1082 Content-Length header, it caused the server to wait for input that libcurl
1083 wouldn't send. Aaron Oneal reported this problem in bug report #2799008
1084 (http://curl.haxx.se/bug/view.cgi?id=2799008) and helped us verify the fix.
1086 Yang Tse (4 Jun 2009)
1087 - Igor Novoseltsev provided patches and information, that after some
1088 adjustments to better fit curl's way of doing things, have resulted
1089 in the posibility of building libcurl for VxWorks.
1091 Daniel Fandrich (2 June 2009)
1092 - Checked in a Google Android make file. To use it, you must first
1093 create a config.h file by running configure in the Android environment,
1094 which doesn't seem to be easy to do. If no easy way can be found, a
1095 static config-android.h may need to be created and checked in to the
1096 libcurl source tree.
1098 Daniel Stenberg (1 June 2009)
1099 - Claes Jakobsson fixed the configure script to better find and use NSS
1102 Yang Tse (1 Jun 2009)
1103 - John E. Malmberg provided a VMS specific clean-up for curl.h, and pointed
1104 out that the configure script was failing to detect the timeval struct on
1105 VMS when building with _XOPEN_SOURCE_EXTENDED undefined due to definition
1106 taking place in socket.h instead of time.h. I have adjusted configure
1107 script to also include this header when checking struct timeval.
1109 Daniel Stenberg (27 May 2009)
1110 - Frank McGeough provided a small OpenSSL #include fix to make libcurl compile
1111 fine with Nokia 5th edition 1.0 SDK for Symbian.
1113 - Andre Guibert de Bruet found a call to a OpenSSL function that didn't check
1114 for a failure properly.
1116 - Mike Crowe pointed out that setting CURLOPT_USERPWD to NULL used to clear
1117 the auth credentials back in 7.19.0 and earlier while now you have to set ""
1118 to get the same effect. His patch brings back the ability to use NULL.
1120 - Claes Jakobsson fixed libcurl-NSS to build fine even without the
1121 PK11_CreateGenericObject() function.
1123 Daniel Stenberg (25 May 2009)
1124 - bug report #2796358 (http://curl.haxx.se/bug/view.cgi?id=2796358) pointed
1125 out that the cookie parser would leak memory when it parses cookies that are
1126 received with domain, path etc set multiple times in the same header. While
1127 such a cookie is questionable, they occur in the wild and libcurl no longer
1128 leaks memory for them. I added such a header to test case 8.
1130 Daniel Fandrich (22 May 2009)
1131 - Removed some obsolete digest code that caused a valgrind error in test 551.
1133 Daniel Fandrich (20 May 2009)
1134 - Added "non-existing host" test keywords to make it easy to skip those
1135 tests on machines that have broken DNS configurations (such as
1136 those configured to use OpenDNS).
1138 Daniel Stenberg (19 May 2009)
1139 - Kamil Dudka brought the patch from the Redhat bug entry
1140 https://bugzilla.redhat.com/show_bug.cgi?id=427966 which was libcurl closing
1141 a bad file descriptor when closing down the FTP data connection. Caolan
1142 McNamara seems to be the original author of it.
1144 Version 7.19.5 (18 May 2009)
1146 Daniel Stenberg (17 May 2009)
1147 - James Bursa posted a patch to the mailing list that fixed a problem with
1148 no_proxy which made it not skip the proxy if the URL entered contained a
1149 user name. I added test case 1101 to verify.
1151 Daniel Stenberg (11 May 2009)
1152 - Balint Szilakszi reported a memory leak when libcurl did gzip decompression
1153 of streams that had some parts (legitimately) missing. We now provide and use
1154 a proper cleanup function for the content encoding submodule.
1155 http://curl.haxx.se/mail/lib-2009-05/0092.html
1157 - Kamil Dudka provided a fix for libcurl-NSS reported by Michael Cronenworth
1158 at https://bugzilla.redhat.com/show_bug.cgi?id=453612#c12
1160 If an incorrect password is given while loading a private key, libcurl ends
1161 up in an infinite loop consuming memory. The bug is critical.
1163 - I fixed the problem with doing NTLM, POST and then following a 302 redirect,
1164 as reported by Ebenezer Ikonne (on curl-users) and Laurent Rabret (on
1165 curl-library). The transfer was mistakenly marked to get more data to send
1166 but since it didn't actually have that, it just hung there...
1168 Daniel Stenberg (10 May 2009)
1169 - Andre Guibert de Bruet correctly pointed out an over-alloc with one wasted
1170 byte in the digest code.
1172 Yang Tse (9 May 2009)
1173 - Removed DOS and TPF package's subdirectory Makefile.am, it was only used
1174 to include some files in the distribution tarball serving no other purpose.
1175 Files from the DOS and TPF subdirectories are now included in the EXTRA_DIST
1176 of the Makefile in the parent subdirectory.
1178 Yang Tse (8 May 2009)
1179 - Changed host name literal in several tests to one under the haxx.se domain.
1181 - Renamed vc6 workspace and project files to avoid filename clash when used
1182 for conversion to later VS versions.
1184 Daniel Stenberg (8 May 2009)
1185 - Constantine Sapuntzakis fixed bug report #2784055
1186 (http://curl.haxx.se/bug/view.cgi?id=2784055) identifying a problem to
1187 connect to SOCKS proxies when using the multi interface. It turned out to
1188 almost not work at all previously. We need to wait for the TCP connect to
1189 be properly verified before doing the SOCKS magic.
1191 There's still a flaw in the FTP code for this.
1193 Daniel Stenberg (7 May 2009)
1194 - Made the SO_SNDBUF setting for the data connection socket for ftp uploads as
1195 well. See change 28 Apr 2009.
1197 Yang Tse (7 May 2009)
1198 - Fixed an issue affecting FTP transfers, introduced with the transfer.c
1199 patch committed May 4.
1201 Daniel Stenberg (7 May 2009)
1202 - Man page *roff problems fixed thanks to input from Colin Watson. Problems
1203 reported in the Debian package.
1205 - Vijay G filed bug report #2723236
1206 (http://curl.haxx.se/bug/view.cgi?id=2723236) identifying a problem with
1207 libcurl's TFTP code and its lack of dealing with the OACK packet.
1209 Yang Tse (5 May 2009)
1210 - Fixed the --ftp-port address of test #251 to the CLIENTIP address, and
1211 reverted the change affecting test suite harness committed 4 May.
1213 Daniel Stenberg (5 May 2009)
1214 - Inspired by Michael Smith's session id fix for OpenSSL, I did the
1215 corresponding fix in the GnuTLS code: make sure to store the new session id
1216 in case the previous re-used one is rejected.
1218 Daniel Stenberg (4 May 2009)
1219 - Michael Smith posted bug report #2786255
1220 (http://curl.haxx.se/bug/view.cgi?id=2786255) with a patch, identifying how
1221 libcurl did not deal with SSL session ids properly if the server rejected a
1222 re-use of one. Starting now, it will forget the rejected one and remember
1223 the new. This change was for OpenSSL only, it is likely that other SSL lib
1224 code needs similar fixes.
1226 Yang Tse (4 May 2009)
1227 - Applied David McCreedy's "transfer.c fixes for CURL_DO_LINEEND_CONV and
1228 non-ASCII platform HTTP requests" patch addressing two HTTP PUT problems:
1229 1) On non-ASCII platforms not all of the protocol portions of the PUT are
1230 being translated to ASCII. 2) On all platforms the line endings of part of
1231 the protocol portions are mangled from CRLF to CRCRLF if data->set.crlf or
1232 data->set.prefer_ascii are set (depending on CURL_DO_LINEEND_CONV).
1234 - Applied David McCreedy's patch to fix test suite harness to allow test FTP
1235 server and client on different machines, providing FTP client address when
1236 running the FTP test server.
1238 Daniel Fandrich (3 May 2009)
1239 - Added and disabled test case 563 which shows KNOWN_BUGS #59. The bug
1240 report failed to mention that a proxy must be used to reproduce it.
1242 Yang Tse (2 May 2009)
1243 - Use a build-time configured curl_socklen_t data type instead of socklen_t.
1245 Yang Tse (1 May 2009)
1246 - Applied David McCreedy's patches "TPF-platform specific changes to various
1247 files" and "http.c fix to Curl_proxyCONNECT for non-ASCII platforms", the
1248 former with minor edits.
1250 Daniel Stenberg (30 Apr 2009)
1251 - I was going to fix issue #59 in KNOWN_BUGS
1253 If the CURLOPT_PORT option is used on an FTP URL like
1254 "ftp://example.com/file;type=A" the ";type=A" is stripped off.
1256 I added test case 562 to verify, only to find out that I couldn't repeat
1257 this bug so I hereby consider it not a bug anymore!
1259 Daniel Stenberg (29 Apr 2009)
1260 - Based on bug report #2723219 (http://curl.haxx.se/bug/view.cgi?id=2723219)
1261 I've now made TFTP "connections" not being kept for re-use within libcurl.
1262 TFTP is UDP-based so the benefit was really low (if even existing) to begin
1263 with so instead of tracking down to fix this problem we instead removed the
1264 re-use. I also enabled test case 1099 that I wrote a few days ago to verify
1265 that this change fixes the reported problem.
1267 Daniel Stenberg (28 Apr 2009)
1268 - Constantine Sapuntzakis filed bug report #2783090
1269 (http://curl.haxx.se/bug/view.cgi?id=2783090) pointing out that on windows
1270 we need to grow the SO_SNDBUF buffer somewhat to get really good upload
1271 speeds. http://support.microsoft.com/kb/823764 has the details. Friends
1272 confirmed that simply adding 32 to CURL_MAX_WRITE_SIZE is enough.
1274 - Bug report #2709004 (http://curl.haxx.se/bug/view.cgi?id=2709004) by Tim
1275 Chen pointed out how curl couldn't upload with resume when reading from a
1278 This ended up with the introduction of a new return code for the
1279 CURLOPT_SEEKFUNCTION callback that basically says that the seek failed but
1280 that libcurl may try to resolve the situation anyway. In our case this means
1281 libcurl will attempt to instead read that much data from the stream instead
1282 of seeking and that way curl can now upload with resume when data is read
1285 Daniel Stenberg (26 Apr 2009)
1286 - Bug report #2779733 (http://curl.haxx.se/bug/view.cgi?id=2779733) by Sven
1287 Wegener pointed out that CURLINFO_APPCONNECT_TIME didn't work with the multi
1288 interface and provided a patch that fixed the problem!
1290 Daniel Stenberg (24 Apr 2009)
1291 - Kamil Dudka fixed another NSS-related leak when client certs were used.
1293 - Bug report #2779245 (http://curl.haxx.se/bug/view.cgi?id=2779245) by Rainer
1294 Koenig pointed out that the man page didn't tell that the *_proxy
1295 environment variables can be specified lower case or UPPER CASE and the
1296 lower case takes precedence,
1298 Daniel Fandrich (21 Apr 2009)
1299 - Added new libcurl source files to Amiga, RiscOS and VC6 build files.
1301 Yang Tse (21 Apr 2009)
1302 - Moved potential inclusion of system's malloc.h and memory.h header files to
1303 setup_once.h. Inclusion of each header file is based on the definition of
1304 NEED_MALLOC_H and NEED_MEMORY_H respectively.
1306 Renamed libcurl's memory.h to curl_memory.h
1308 Daniel Stenberg (20 Apr 2009)
1309 - Leanic Lefever reported a crash and did some detailed research on why and
1310 how it occurs (http://curl.haxx.se/mail/lib-2009-04/0289.html). The
1311 conclusion was that if an error is detected and Curl_done() is called for
1312 the connection, ftp_done() could at times return another error code that
1313 then would take precedence and that new code confused existing logic that
1314 works for the first error code (CURLE_SEND_ERROR) only.
1316 - Gisle Vanem noticed that --libtool would produce bogus strings at times for
1317 OBJECTPOINT options. Now we've introduced a new function - my_setopt_str -
1318 within the app for setting plain string options to avoid the risk of this
1321 Daniel Stenberg (17 Apr 2009)
1322 - Pramod Sharma reported and tracked down a bug when doing FTP over a HTTP
1323 proxy. libcurl would then wrongly close the connection after each
1324 request. In his case it had the weird side-effect that it killed NTLM auth
1325 for the proxy causing an inifinite loop!
1327 I added test case 1098 to verify this fix. The test case does however not
1328 properly verify that the transfers are done persistently - as I couldn't
1329 think of a clever way to achieve it right now - but you need to read the
1330 stderr output after a test run to see that it truly did the right thing.
1332 Daniel Stenberg (13 Apr 2009)
1333 - bug report #2727981 (http://curl.haxx.se/bug/view.cgi?id=2727981) by Martin
1334 Storsjö pointed out how setting CURLOPT_NOBODY to 0 could be downright
1335 confusing as it set the method to either GET or HEAD. The example he showed
1338 curl_easy_setopt(curl, CURLOPT_PUT, 1);
1339 curl_easy_setopt(curl, CURLOPT_NOBODY, 0);
1341 The new way doesn't alter the method until the request is about to start. If
1342 CURLOPT_NOBODY is then 1 the HTTP request will be HEAD. If CURLOPT_NOBODY is
1343 0 and the request happens to have been set to HEAD, it will then instead be
1344 set to GET. I believe this will be less surprising to users, and hopefully
1345 not hit any existing users badly.
1347 - Toshio Kuratomi reported a memory leak problem with libcurl+NSS that turned
1348 out to be leaking cacerts. Kamil Dudka helped me complete the fix. The issue
1349 is found in Redhat's bug tracker:
1350 https://bugzilla.redhat.com/show_bug.cgi?id=453612
1352 There are still memory leaks present, but they seem to have other reasons.
1354 Daniel Fandrich (11 Apr 2009)
1355 - Added new libcurl source files to Symbian OS build files.
1356 - Improved Symbian support for SSL.
1358 Yang Tse (10 Apr 2009)
1359 - Daniel Johnson improved the MacOSX-Framework shell script to now perform all
1360 the steps required to build a Mac OS X four way fat ppc/i386/ppc64/x86_64
1361 libcurl.framework. Four way fat framework requires OS X 10.5 SDK or later.
1363 Yang Tse (8 Apr 2009)
1364 - Removed Sun compilers preprocessor block from curlbuild.h.dist, this also
1365 removes it from the curlbuild.h file originally distributed by the cURL
1366 project as this file is intended for systems not capable of running the
1367 configure script. For those who have been building curl out of the source
1368 code curl distribution tarball provided by curl.haxx.se the change implies
1369 nothing. Previous change in this area committed 2 Apr becomes irrelevant.
1371 Daniel Stenberg (6 Apr 2009)
1372 - I clarified in the docs that CURLOPT_SEEKFUNCTION should return 0 on success
1373 and 1 on fatal errors. Previously it only mentioned non-zero on fatal
1374 errors. This is a slight change in meaning, but it follows what we've done
1375 elsewhere before and it opens up for LOTS of more useful return codes
1376 whenever we can think of them...
1378 Yang Tse (2 Apr 2009)
1379 - Fix curl_off_t definition for builds done using Sun compilers and a
1380 non-configured libcurl. In this case curl_off_t data type was gated
1381 to the off_t data type which depends on the _FILE_OFFSET_BITS. This
1382 configuration is exactly the unwanted configuration for our curl_off_t
1383 data type which must not depend on such setting. This breaks ABI for
1384 libcurl libraries built with Sun compilers which were built without
1385 having run the configure script with _FILE_OFFSET_BITS different than
1386 64 and using the ILP32 data model.
1388 Daniel Stenberg (1 Apr 2009)
1389 - Andre Guibert de Bruet fixed a NULL pointer use in an infof() call if a
1390 strdup() call failed.
1392 Daniel Fandrich (31 Mar 2009)
1393 - Properly return an error code in curl_easy_recv (reported by Jim Freeman).
1395 Daniel Stenberg (18 Mar 2009)
1396 - Kamil Dudka brought a patch that enables 6 additional crypto algorithms when
1397 NSS is used. These ciphers were added in NSS 3.4 and require to be enabled
1400 Daniel Stenberg (13 Mar 2009)
1401 - Use libssh2_version() to present the libssh2 version in case the libssh2
1402 library is found to support it.
1404 Yang Tse (12 Mar 2009)
1405 - Added missing Curl_read() return code checking in TELNET transfers.
1407 - Pierre Brico found and fixed TELNET transfers not being aborted upon
1408 a write callback failure.
1410 Daniel Stenberg (11 Mar 2009)
1411 - Kamil Dudka made the curl tool properly call curl_global_init() before any
1412 other libcurl function.
1414 Yang Tse (11 Mar 2009)
1415 - Added missing TELNET timeout support for Windows builds. This issue was
1416 reported by Pierre Brico.
1418 Daniel Stenberg (9 Mar 2009)
1419 - Frank Hempel found out a bug and provided the fix:
1421 curl_easy_duphandle did not necessarily duplicate the CURLOPT_COOKIEFILE
1422 option. It only enabled the cookie engine in the destination handle if
1423 data->cookies is not NULL (where data is the source handle). In case of a
1424 newly initialized handle which just had the cookie support enabled by a
1425 curl_easy_setopt(handle, CURL_COOKIEFILE, "")-call, handle->cookies was
1426 still NULL because the setopt-call only appends the value to
1427 data->change.cookielist, hence duplicating this handle would not have the
1428 cookie engine switched on.
1430 We also concluded that the slist-functionality would be suitable for being
1431 put in its own module rather than simply hanging out in lib/sendf.c so I
1432 created lib/slist.[ch] for them.
1434 - Andreas Farber made the 'buildconf' script check for the presence of m4
1435 scripts to make it detect a bad checkout earlier. People with older
1436 checkouts who don't do cvs update with the -d option won't get the new dirs
1437 and then will get funny outputs that can be a bit hard to understand and
1440 Daniel Stenberg (8 Mar 2009)
1441 - Andre Guibert de Bruet found and fixed a code segment in ssluse.c where the
1442 allocation of the memory BIO was not being properly checked.
1444 - Andre Guibert de Bruet fixed the gnutls-using code: There are a few places
1445 in the gnutls code where we were checking for negative values for errors,
1446 when the man pages state that GNUTLS_E_SUCCESS is returned on success and
1447 other values indicate error conditions.
1449 - Bill Egert pointed out (http://curl.haxx.se/bug/view.cgi?id=2671602) that
1450 curl didn't use sprintf() in a way that is documented to work in POSIX but
1451 since we use our own printf() code (from libcurl) that shouldn't be a
1452 problem. Nonetheless I modified the code to not rely on such particular
1453 features and to not cause further raised eyebrowse with no good reason.
1455 Daniel Fandrich (5 Mar 2009)
1456 - Expanded the security section of the libcurl-tutorial man page to cover
1457 more issues for authors to consider when writing robust libcurl-using
1460 Yang Tse (5 Mar 2009)
1461 - Fixed NTLM authentication memory leak on SSPI enabled Windows builds. This
1462 issue was noticed by Chris Deidun.
1464 Daniel Fandrich (4 Mar 2009)
1465 - Fixed a problem with m4 quoting in the OpenSSL configure check reported
1468 Daniel Stenberg (3 Mar 2009)
1469 - David James brought a patch that make libcurl close (all) dead connections
1470 whenever you attempt to open a new connection.
1472 1. After cleaning up a dead connection, "continue" instead of
1473 returning FALSE. This ensures that we clean up all dead connections,
1474 rather than just cleaning up the first dead connection.
1475 2. Move up the cleanup for dead connections so that it occurs for
1476 all connections, rather than just the connections which have the same
1477 preferences as our current new connection.
1479 Version 7.19.4 (3 March 2009)
1481 Daniel Stenberg (3 Mar 2009)
1482 - David Kierznowski notified us about a security flaw
1483 (http://curl.haxx.se/docs/adv_20090303.html also known as CVE-2009-0037) in
1484 which previous libcurl versions (by design) can be tricked to access an
1485 arbitrary local/different file instead of a remote one when
1486 CURLOPT_FOLLOWLOCATION is enabled. This flaw is now fixed in this release
1487 together this the addition of two new setopt options for controlling this
1490 o CURLOPT_REDIR_PROTOCOLS controls what protocols libcurl is allowed to
1491 follow to when CURLOPT_FOLLOWLOCATION is enabled. By default, this option
1492 excludes the FILE and SCP protocols and thus you nee to explicitly allow
1493 them in your app if you really want that behavior.
1495 o CURLOPT_PROTOCOLS controls what protocol(s) libcurl is allowed to fetch
1496 using the primary URL option. This is useful if you want to allow a user or
1497 other outsiders control what URL to pass to libcurl and yet not allow all
1498 protocols libcurl may have been built to support.
1500 Daniel Stenberg (27 Feb 2009)
1501 - Senthil Raja Velu reported a problem when CURLOPT_INTERFACE and
1502 CURLOPT_LOCALPORT were used together (the local port bind failed), and
1503 Markus Koetter provided the fix!
1505 Daniel Stenberg (25 Feb 2009)
1506 - As Daniel Fandrich figured out, we must do the GnuTLS initing in the
1507 curl_global_init() function to properly maintain the performing functions
1508 thread-safe. We've previously (28 April 2007) moved the init to a later time
1509 just to avoid it to fail very early when libgcrypt dislikes the situation,
1510 but that move was bad and the fix should rather be in libgcrypt or
1513 Daniel Stenberg (24 Feb 2009)
1514 - Brian J. Murrell found out that Negotiate proxy authentication didn't work.
1515 It happened because the code used the struct for server-based auth all the
1516 time for both proxy and server auth which of course was wrong.
1518 Daniel Stenberg (23 Feb 2009)
1519 - After a bug reported by James Cheng I've made curl_easy_getinfo() for
1520 CURLINFO_CONTENT_LENGTH_DOWNLOAD and CURLINFO_CONTENT_LENGTH_UPLOAD return
1521 -1 if the sizes aren't know. Previously these returned 0, make it impossible
1522 to detect the difference between actually zero and unknown.
1524 Yang Tse (23 Feb 2009)
1525 - Daniel Johnson provided a shell script that will perform all the steps needed
1526 to build a Mac OS X fat ppc/i386 or ppc64/x86_64 libcurl.framework
1528 Daniel Stenberg (23 Feb 2009)
1529 - I renamed everything in the windows builds files that used the name 'curllib'
1530 to the proper 'libcurl' as clearly this caused confusion.
1532 Yang Tse (20 Feb 2009)
1533 - Do not halt compilation when using VS2008 to build a Windows 2000 target.
1535 Daniel Stenberg (20 Feb 2009)
1536 - Linus Nielsen Feltzing reported and helped me repeat and fix a problem with
1537 FTP with the multi interface: when a transfer fails, like when aborted by a
1538 write callback, the control connection was wrongly closed and thus not
1541 This change is also an attempt to cleanup the code somewhat in this area, as
1542 now the FTP code attempts to keep (better) track on pending responses
1543 necessary to get read in ftp_done().
1545 Daniel Stenberg (19 Feb 2009)
1546 - Patrik Thunstrom reported a problem and helped me repeat it. It turned out
1547 libcurl did a superfluous 1000ms wait when doing SFTP downloads!
1549 We read data with libssh2 while doing the "DO" operation for SFTP and then
1550 when we were about to start getting data for the actual file part, the
1551 "TRANSFER" part, we waited for socket action (in 1000ms) before doing a
1552 libssh2-read. But in this case libssh2 had already read and buffered the
1553 data so we ended up always just waiting 1000ms before we get working on the
1556 Patrick Monnerat (18 Feb 2009)
1557 - FTP downloads (i.e.: RETR) ending with code 550 now return error
1558 CURLE_REMOTE_FILE_NOT_FOUND instead of CURLE_FTP_COULDNT_RETR_FILE.
1560 Daniel Stenberg (17 Feb 2009)
1561 - Kamil Dudka made NSS-powered builds compile and run again!
1563 - A second follow-up change by Andre Guibert de Bruet to fix a related memory
1564 leak like that fixed on the 14th. When zlib returns failure, we need to
1565 cleanup properly before returning error.
1567 - CURLOPT_FTP_CREATE_MISSING_DIRS can now be set to 2 in addition to 1 for
1568 plain FTP connections, and it will then allow MKD to fail once and retry the
1569 CWD afterwards. This is especially useful if you're doing many simultanoes
1570 connections against the same server and they all have this option enabled,
1571 as then CWD may first fail but then another connection does MKD before this
1572 connection and thus MKD fails but trying CWD works! The numbers can
1573 (should?) now be set with the convenience enums now called
1574 CURLFTP_CREATE_DIR and CURLFTP_CREATE_DIR_RETRY.
1576 Tests has proven that if you're making an application that uploads a set of
1577 files to an ftp server, you will get a noticable gain in speed if you're
1578 using multiple connections and this option will be then be very useful.
1580 Daniel Stenberg (14 Feb 2009)
1581 - Andre Guibert de Bruet found and fixed a memory leak in the content encoding
1582 code, which could happen on libz errors.
1584 Daniel Fandrich (12 Feb 2009)
1585 - Added support for Digest and NTLM authentication using GnuTLS.
1587 Daniel Stenberg (11 Feb 2009)
1588 - CURLINFO_CONDITION_UNMET was added to allow an application to get to know if
1589 the condition in the previous request was unmet. This is typically a time
1590 condition set with CURLOPT_TIMECONDITION and was previously not possible to
1591 reliably figure out. From bug report #2565128
1592 (http://curl.haxx.se/bug/view.cgi?id=2565128) filed by Jocelyn Jaubert.
1594 Daniel Fandrich (4 Feb 2009)
1595 - Don't add the standard /usr/lib or /usr/include paths to LDFLAGS and CPPFLAGS
1596 (respectively) when --with-ssl=/usr is used (patch based on FreeBSD).
1598 - Added an explicit buffer limit check in msdosify() (patch based on FreeBSD).
1599 This couldn't ever overflow in curl, but might if the code were used
1600 elsewhere or under different conditions.
1602 Daniel Stenberg (3 Feb 2009)
1603 - Hidemoto Nakada provided a small fix that makes it possible to get the
1604 CURLINFO_CONTENT_LENGTH_DOWNLOAD size from file:// "transfers" with
1605 CURLOPT_NOBODY set true.
1607 Daniel Stenberg (2 Feb 2009)
1608 - Patrick Scott found a rather large memory leak when using the multi
1609 interface and setting CURLMOPT_MAXCONNECTS to something less than the number
1610 of handles you add to the multi handle. All the connections that didn't fit
1611 in the cache would not be properly disconnected nor freed!
1613 - Craig A West brought us: libcurl now defaults to do CONNECT with HTTP
1614 version 1.1 instead of 1.0 like before. This change also introduces the new
1615 proxy type for libcurl called 'CURLPROXY_HTTP_1_0' that then allows apps to
1616 switch (back) to CONNECT 1.0 requests. The curl tool also got a --proxy1.0
1617 option that works exactly like --proxy but sets CURLPROXY_HTTP_1_0.
1619 I updated all test cases cases that use CONNECT and I tried to do some using
1620 --proxy1.0 and some updated to do CONNECT 1.1 to get both versions run.
1622 Daniel Stenberg (31 Jan 2009)
1623 - When building with c-ares 1.6.1 (not yet released) or later and IPv6 support
1624 enabled, we can now take advantage of its brand new AF_UNSPEC support in
1625 ares_gethostbyname(). This makes test case 241 finally run fine for me with
1626 this setup since it now parses the "::1 ip6-localhost" line fine in my
1629 Daniel Stenberg (30 Jan 2009)
1630 - Scott Cantor filed bug report #2550061
1631 (http://curl.haxx.se/bug/view.cgi?id=2550061) mentioning that I failed to
1632 properly make sure that the VC9 makefiles got included in the latest
1633 release. I've now fixed the release script and verified it so next release
1634 will hopefully include them properly!
1636 Daniel Fandrich (30 Jan 2009)
1637 - Fixed --disable-proxy for FTP and SOCKS. Thanks to Daniel Egger for
1640 Yang Tse (29 Jan 2009)
1641 - Introduced curl_sspi.c and curl_sspi.h for the implementation of functions
1642 Curl_sspi_global_init() and Curl_sspi_global_cleanup() which previously were
1643 named Curl_ntlm_global_init() and Curl_ntlm_global_cleanup() in http_ntlm.c
1644 Also adjusted socks_sspi.c to remove the link-time dependency on the Windows
1645 SSPI library using it now in the same way as it was done in http_ntlm.c.
1647 Daniel Stenberg (28 Jan 2009)
1648 - Markus Moeller introduced two new options to libcurl:
1649 CURLOPT_SOCKS5_GSSAPI_SERVICE and CURLOPT_SOCKS5_GSSAPI_NEC to allow libcurl
1650 to do GSS-style authentication with SOCKS5 proxies. The curl tool got the
1651 options called --socks5-gssapi-service and --socks5-gssapi-nec to enable
1654 Daniel Stenberg (26 Jan 2009)
1655 - Chad Monroe provided the new CURLOPT_TFTP_BLKSIZE option that allows an app
1656 to set desired block size to use for TFTP transfers instead of the default
1659 - The "-no_ticket" option was introduced in Openssl0.9.8j. It's a flag to
1660 disable "rfc4507bis session ticket support". rfc4507bis was later turned
1661 into the proper RFC5077 it seems: http://tools.ietf.org/html/rfc5077
1663 The enabled extension concerns the session management. I wonder how often
1664 libcurl stops a connection and then resumes a TLS session. also, sending the
1665 session data is some overhead. .I suggest that you just use your proposed
1666 patch (which explicitly disables TICKET).
1668 If someone writes an application with libcurl and openssl who wants to
1669 enable the feature, one can do this in the SSL callback.
1671 Sharad Gupta brought this to my attention. Peter Sylvester helped me decide
1672 on the proper action.
1674 - Alexey Borzov filed bug report #2535504
1675 (http://curl.haxx.se/bug/view.cgi?id=2535504) pointing out that realms with
1676 quoted quotation marks in HTTP Digest headers didn't work. I've now added
1677 test case 1095 that verifies my fix.
1679 - Craig A West brought CURLOPT_NOPROXY and the corresponding --noproxy option.
1680 They basically offer the same thing the NO_PROXY environment variable only
1681 offered previously: list a set of host names that shall not use the proxy
1682 even if one is specified.
1684 Daniel Fandrich (20 Jan 2009)
1685 - Call setlocale() for libtest tests to test the effects of locale-induced
1686 libc changes on libcurl.
1688 - Fixed a couple more locale-dependent toupper conversions, mainly for
1689 clarity. This does fix one problem that causes ;type=i FTP URLs
1690 to fail in the Turkish locale when CURLOPT_PROXY_TRANSFER_MODE is
1691 used (test case 561)
1693 - Added tests 561 and 1091 through 1094 to test various combinations
1694 of ;type= and ;mode= URLs that could potentially fail in the Turkish
1697 Daniel Stenberg (20 Jan 2009)
1698 - Lisa Xu pointed out that the ssh.obj file was missing from the
1699 lib/Makefile.vc6 file (and thus from the vc8 and vc9 ones too).
1701 Version 7.19.3 (19 January 2009)
1703 Daniel Stenberg (16 Jan 2009)
1704 - Andrew de los Reyes fixed curlbuild.h for "generic" gcc builds on PPC, both
1707 Daniel Stenberg (15 Jan 2009)
1708 - Tim Ansell fixed a compiler warning in lib/cookie.c
1710 Daniel Stenberg (14 Jan 2009)
1711 - Grant Erickson fixed timeouts for TFTP such that specifying a
1712 connect-timeout, a max-time or both options work correctly and as expected
1713 by passing the correct boolean value to Curl_timeleft via the
1714 'duringconnect' parameter.
1716 With this small change, curl TFTP now behaves as expected (and likely as
1717 originally-designed):
1719 1) For non-existent or unreachable dotted IP addresses:
1721 a) With no options, follows the default curl 300s timeout...
1722 b) With --connect-timeout only, follows that value...
1723 c) With --max-time only, follows that value...
1724 d) With both --connect-timeout and --max-time, follows the smaller value...
1726 and times out with a "curl: (7) Couldn't connect to server" error.
1728 2) For transfers to/from a valid host:
1730 a) With no options, follows default curl 300s timeout for the
1731 first XRQ/DATA/ACK transaction and the default TFTP 3600s
1732 timeout for the remainder of the transfer...
1734 b) With --connect-time only, follows that value for the
1735 first XRQ/DATA/ACK transaction and the default TFTP 3600s
1736 timeout for the remainder of the transfer...
1738 c) With --max-time only, follows that value for the first
1739 XRQ/DATA/ACK transaction and for the remainder of the
1742 d) With both --connect-timeout and --max-time, follows the former
1743 for the first XRQ/DATA/ACK transaction and the latter for the
1744 remainder of the transfer...
1746 and times out with a "curl: (28) Timeout was reached" error as
1749 Daniel Stenberg (13 Jan 2009)
1750 - Michael Wallner fixed a NULL pointer deref when calling
1751 curl_easy_setup(curl, CURLOPT_COOKIELIST, "SESS") on a CURL handle with no
1754 - Stefan Teleman brought a patch to fix the default curlbuild.h file for the
1757 Daniel Stenberg (12 Jan 2009)
1758 - Based on bug report #2498665 (http://curl.haxx.se/bug/view.cgi?id=2498665)
1759 by Daniel Black, I've now added magic to the configure script that makes it
1760 use pkg-config to detect gnutls details as well if the existing method
1761 (using libgnutls-config) fails. While doing this, I cleaned up and unified
1762 the pkg-config usage when detecting openssl and nss as well.
1764 Daniel Stenberg (11 Jan 2009)
1765 - Karl Moerder brought the patch that creates vc9 Makefiles, and I made
1766 'maketgz' now use the actual makefile targets to do the VC8 and VC9
1769 Daniel Stenberg (10 Jan 2009)
1770 - Emil Romanus fixed:
1772 When using the multi interface over HTTP and the server returns a Location
1773 header, the running easy handle will get stuck in the CURLM_STATE_PERFORM
1774 state, leaving the external event loop stuck waiting for data from the
1775 ingoing socket (when using the curl_multi_socket_action stuff). While this
1776 bug was pretty hard to find, it seems to require only a one-line fix. The
1777 break statement on line 1374 in multi.c caused the function to skip the call
1780 How to reproduce this bug? Well, that's another question. evhiperfifo.c in
1781 the examples directory chokes on this bug only _sometimes_, probably
1782 depending on how fast the URLs are added. One way of testing the bug out is
1783 writing to hiper.fifo from more than one source at the same time.
1785 Daniel Fandrich (7 Jan 2009)
1786 - Unified much of the SessionHandle initialization done in Curl_open() and
1787 curl_easy_reset() by creating Curl_init_userdefined(). This had the side
1788 effect of fixing curl_easy_reset() so it now also resets
1789 CURLOPT_FTP_FILEMETHOD and CURLOPT_SSL_SESSIONID_CACHE
1791 Daniel Stenberg (7 Jan 2009)
1792 - Rob Crittenden did once again provide an NSS update:
1794 I have to jump through a few hoops now with the NSS library initialization
1795 since another part of an application may have already initialized NSS by the
1796 time Curl gets invoked. This patch is more careful to only shutdown the NSS
1797 library if Curl did the initialization.
1799 It also adds in a bit of code to set the default ciphers if the app that
1800 call NSS_Init* did not call NSS_SetDomesticPolicy() or set specific
1801 ciphers. One might argue that this lets other application developers get
1802 lazy and/or they aren't using the NSS API correctly, and you'd be right.
1803 But still, this will avoid terribly difficult-to-trace crashes and is
1806 Daniel Stenberg (1 Jan 2009)
1807 - 'reconf' is removed since we rather have users use 'buildconf'
1809 Daniel Stenberg (31 Dec 2008)
1810 - Bas Mevissen reported http://curl.haxx.se/bug/view.cgi?id=2479030 pointing
1811 out that 'reconf' didn't properly point out the m4 subdirectory when running
1814 Daniel Stenberg (29 Dec 2008)
1815 - Phil Lisiecki filed bug report #2413067
1816 (http://curl.haxx.se/bug/view.cgi?id=2413067) that identified a problem that
1817 would cause libcurl to mark a DNS cache entry "in use" eternally if the
1818 subsequence TCP connect failed. It would thus never get pruned and refreshed
1819 as it should've been.
1821 Phil provided his own patch to this problem that while it seemed to work
1822 wasn't complete and thus I wrote my own fix to the problem.
1824 Daniel Stenberg (28 Dec 2008)
1825 - Peter Korsgaard fixed building libcurl with "configure --with-ssl
1828 - Anthony Bryan fixed more language and spelling flaws in man pages.
1830 Daniel Stenberg (22 Dec 2008)
1831 - Given a recent enough libssh2, libcurl can now seek/resume with SFTP even
1832 on file indexes beyond 2 or 4GB.
1834 - Anthony Bryan provided a set of patches that cleaned up manual language,
1835 corrected spellings and more.
1837 Daniel Stenberg (20 Dec 2008)
1838 - Igor Novoseltsev fixed a bad situation for the multi_socket() API when doing
1839 pipelining, as libcurl could then easily get confused and A) work on the
1840 handle that was not "first in queue" on a pipeline, or even B) tell the app
1841 to REMOVE a socket while it was in use by a second handle in a pipeline. Both
1842 errors caused hanging or stalling applications.
1844 Daniel Stenberg (19 Dec 2008)
1845 - curl_multi_timeout() could return a timeout value of 0 even though nothing
1846 was actually ready to get done, as the internal time resolution is higher
1847 than the returned millisecond timer. Therefore it could cause applications
1848 running on fast processors to do short bursts of busy-loops.
1849 curl_multi_timeout() will now only return 0 if the timeout is actually
1852 - Using the libssh2 0.19 function libssh2_session_block_directions(), libcurl
1853 now has an improved ability to do right when the multi interface (both
1854 "regular" and multi_socket) is used for SCP and SFTP transfers. This should
1855 result in (much) less busy-loop situations and thus less CPU usage with no
1858 Daniel Stenberg (17 Dec 2008)
1859 - SCP and SFTP with the multi interface had the same flaw: the 'DONE'
1860 operation didn't complete properly if the EAGAIN equivalent was returned but
1861 libcurl would simply continue with a half-completed close operation
1862 performed. This ruined persistent connection re-use and cause some
1863 SSH-protocol errors in general. The correction is unfortunately adding a
1864 blocking function - doing it entirely non-blocking should be considered for
1867 Gisle Vanem (16 Dec 2008)
1868 - Added the possibility to use the Watt-32 tcp/ip stack under Windows.
1869 The change simply involved adding a USE_WATT32 section in the
1870 config-win32.h files (under ./lib and ./src). This section disables
1871 the use of any Winsock headers.
1873 Daniel Stenberg (16 Dec 2008)
1874 - libssh2_sftp_last_error() was wrongly used at some places in libcurl which
1875 made libcurl sometimes not properly abort problematic SFTP transfers.
1877 Daniel Stenberg (12 Dec 2008)
1878 - More work with Igor Novoseltsev to first fix the remaining stuff for
1879 removing easy handles from multi handles when the easy handle is/was within
1880 a HTTP pipeline. His bug report #2351653
1881 (http://curl.haxx.se/bug/view.cgi?id=2351653) was also related and was
1882 eventually fixed by a patch by Igor himself.
1884 Yang Tse (12 Dec 2008)
1885 - Patrick Monnerat fixed a build regression, introduced in 7.19.2, affecting
1886 OS/400 compilations with IPv6 enabled.
1888 Daniel Stenberg (12 Dec 2008)
1889 - Mark Karpeles filed bug report #2416182 titled "crash in ConnectionExists
1890 when using duphandle+curl_mutli"
1891 (http://curl.haxx.se/bug/view.cgi?id=2416182) which showed that
1892 curl_easy_duphandle() wrongly also copied the pointer to the connection
1893 cache, which was plain wrong and caused a segfault if the handle would be
1894 used in a different multi handle than the handle it was duplicated from.
1896 Daniel Stenberg (11 Dec 2008)
1897 - Keshav Krity found out that libcurl failed to deal with dotted IPv6
1898 addresses if they were very long (>39 letters) due to a too strict address
1899 validity parser. It now accepts addresses up to 45 bytes long.
1901 Daniel Stenberg (11 Dec 2008)
1902 - Internet Explorer had a broken HTTP digest authentication before v7 and
1903 there are servers "out there" that relies on the client doing this broken
1904 Digest authentication. Apache even comes with an option to work with such
1907 The difference is only for URLs that contain a query-part (a '?'-letter and
1908 text to the right of it).
1910 libcurl now supports this quirk, and you enable it by setting the
1911 CURLAUTH_DIGEST_IE bit in the bitmask you pass to the CURLOPT_HTTPAUTH or
1912 CURLOPT_PROXYAUTH options. They are thus individually controlled to server
1915 (note that there's no way to activate this with the curl tool yet)
1917 Daniel Fandrich (9 Dec 2008)
1918 - Added test cases 1089 and 1090 to test --write-out after a redirect to
1919 test a report that the size didn't work, but these test cases pass.
1921 - Documented CURLOPT_CONNECT_ONLY as being useful only on HTTP URLs.
1923 Daniel Stenberg (9 Dec 2008)
1924 - Ken Hirsch simplified how libcurl does FTPS: now it doesn't assume any
1925 particular state for the control connection like it did before for implicit
1926 FTPS (libcurl assumed such control connections to be encrypted while some
1927 FTPS servers such as FileZilla assumes such connections to be clear
1928 mode). Use the CURLOPT_USE_SSL option to set your desired level.
1930 Daniel Stenberg (8 Dec 2008)
1931 - Fred Machado posted about a weird FTP problem on the curl-users list and when
1932 researching it, it turned out he got a 550 response back from a SIZE command
1933 and then I fell over the text in RFC3659 that says:
1935 The presence of the 550 error response to a SIZE command MUST NOT be taken
1936 by the client as an indication that the file cannot be transferred in the
1937 current MODE and TYPE.
1939 In other words: the change I did on September 30th 2008 and that has been
1940 included in the last two releases were a regression and a bad idea. We MUST
1941 NOT take a 550 response from SIZE as a hint that the file doesn't exist.
1943 - Christian Krause filed bug #2221237
1944 (http://curl.haxx.se/bug/view.cgi?id=2221237) that identified an infinite
1945 loop during GSS authentication given some specific conditions. With his
1946 patience and great feedback I managed to narrow down the problem and
1947 eventually fix it although I can't test any of this myself!
1949 Daniel Fandrich (3 Dec 2008)
1950 - Fixed the getifaddrs version of Curl_if2ip to work on systems without IPv6
1951 support (e.g. Minix)
1953 Daniel Stenberg (3 Dec 2008)
1954 - Igor Novoseltsev filed bug #2351645
1955 (http://curl.haxx.se/bug/view.cgi?id=2351645) that identified a problem with
1956 the multi interface that occured if you removed an easy handle while in
1957 progress and the handle was used in a HTTP pipeline.
1959 - Pawel Kierski pointed out a mistake in the cookie code that could lead to a
1960 bad fclose() after a fatal error had occured.
1961 (http://curl.haxx.se/bug/view.cgi?id=2382219)
1963 Daniel Fandrich (25 Nov 2008)
1964 - If a HTTP request is Basic and num is already >=1000, the HTTP test
1965 server adds 1 to num to get the data section to return. This allows
1966 testing authentication negotiations using the Basic authentication
1969 - Added tests 1087 and 1088 to test Basic authentication on a redirect
1970 with and without --location-trusted
1972 Daniel Stenberg (24 Nov 2008)
1973 - Based on a patch by Vlad Grachov, libcurl now uses a new libssh2 0.19
1974 function when built to support SCP and SFTP that helps the library to know
1975 in which direction a particular libssh2 operation would return EAGAIN so
1976 that libcurl knows what socket conditions to wait for before trying the
1977 function call again. Previously (and still when using libssh2 0.18 or
1978 earlier), libcurl will busy-loop in this situation when the easy interface
1981 Daniel Fandrich (20 Nov 2008)
1982 - Automatically detect OpenBSD's CA cert bundle.
1984 Daniel Stenberg (19 Nov 2008)
1985 - I removed the default use of "Pragma: no-cache" from libcurl when a proxy is
1986 used. It has been used since forever but it was never a good idea to use
1987 unless explicitly asked for.
1989 - Josef Wolf's extension that allows a $TESTDIR/gdbinit$testnum file that when
1990 you use runtests.pl -g, will be sourced by gdb to allow additional fancy or
1991 whatever you see fit
1993 - Christian Krause reported and fixed a memory leak that would occur with HTTP
1994 GSS/kerberos authentication (http://curl.haxx.se/bug/view.cgi?id=2284386)
1996 - Andreas Wurf and Markus Koetter helped me analyze a problem that Andreas got
1997 when uploading files to a single FTP server using multiple easy handle
1998 handles with the multi interface. Occasionally a handle would stall in
2001 The problem turned out to be a side-effect of the ConnectionExists()
2002 function's eagerness to re-use a handle for HTTP pipelining so it would
2003 select it even if already being in use, due to an inadequate check for its
2004 chances of being used for pipelnining.
2006 Daniel Fandrich (17 Nov 2008)
2007 - Added more compiler warning options for gcc 4.3
2009 Yang Tse (17 Nov 2008)
2010 - Fix a remaining problem in the inet_pton() runtime configure check. And
2011 fix internal Curl_inet_pton() failures to reject certain malformed literals.
2013 - Make configure script check if ioctl with the SIOCGIFADDR command can be
2014 used, and define HAVE_IOCTL_SIOCGIFADDR if appropriate.
2016 Daniel Stenberg (16 Nov 2008)
2017 - Christian Krause fixed a build failure when building with gss support
2018 enabled and FTP disabled.
2020 - Added check for NULL returns from strdup() in src/main.c and lib/formdata.c
2021 - reported by Jim Meyering also prevent buffer overflow on MSDOS when you do
2022 for example -O on a url with a file name part longer than PATH_MAX letters
2024 - lib/nss.c fixes based on the report by Jim Meyering: I went over and added
2025 checks for return codes for all calls to malloc and strdup that were
2026 missing. I also changed a few malloc(13) to use arrays on the stack and a
2027 few malloc(PATH_MAX) to instead use aprintf() to lower memory use.
2029 - I fixed a memory leak in Curl_nss_connect() when CURLOPT_ISSUERCERT is
2032 Daniel Fandrich (14 Nov 2008)
2033 - Added .xml as one of the few common file extensions known by the multipart
2036 - Added some #ifdefs around header files and change the EAGAIN test to
2037 fix compilation on Cell (reported by Jeff Curley).
2039 Yang Tse (14 Nov 2008)
2040 - Fixed several configure script issues affecting checks for inet_ntoa_r(),
2041 inet_ntop(), inet_pton(), getifaddrs(), fcntl() and getaddrinfo().
2043 Yang Tse (13 Nov 2008)
2044 - Refactored configure script detection of functions used to set sockets into
2045 non-blocking mode, and decouple function detection from function capability.