5 \___|\___/|_| \_\_____|
9 Daniel Stenberg (20 Feb 2010)
10 - I made the SMTP code expect a 250 response back from the server after the
11 full DATA has been sent, and I modified the test SMTP server to also send
12 that response. As usual, the DONE operation that is made after a completed
13 transfer is still not doable in a non-blocking way so this waiting for 250
14 is unfortunately made blockingly.
16 Yang Tse (14 Feb 2010)
17 - Overhauled test suite getpart() function. Fixing potential out of bounds
18 stack and memory overwrites triggered with huge test case definitions.
20 Daniel Stenberg (13 Feb 2010)
21 - Martin Hager reported and fixed a problem with a missing quote in libcurl.m4
23 (http://curl.haxx.se/bug/view.cgi?id=2951319)
25 - Tom Donovan fixed the CURL_FORMAT_* defines when building with cmake.
27 (http://curl.haxx.se/bug/view.cgi?id=2951269)
29 Daniel Stenberg (12 Feb 2010)
30 - Jack Zhang reported a problem with SMTP: we wrongly used multiple addresses
31 in the same RCPT TO line, when they should be sent in separate single
32 commands. I updated test case 802 to verify this.
34 - I also fixed a bad use of my_setopt_str() of CURLOPT_MAIL_RCPT in the curl
35 tool which made it try to output it as string for the --libcurl feature
36 which could lead to crashes.
38 Yang Tse (11 Feb 2010)
39 - Steven M. Schweda fixed VMS builder bad behavior when used in a batch job,
40 removed obsolete batch_compile.com and defines.com and updated VMS readme.
42 Version 7.20.0 (9 February 2010)
44 Daniel Stenberg (9 Feb 2010)
45 - When downloading compressed content over HTTP and the app asked libcurl to
46 automatically uncompress it with the CURLOPT_ENCODING option, libcurl could
47 wrongly provide the callback with more data than the maximum documented
48 amount. An application could thus get tricked into badness if the maximum
49 limit was trusted to be enforced by libcurl itself (as it is documented).
51 This is further detailed and explained in the libcurl security advisory
54 http://curl.haxx.se/docs/adv_20100209.html
56 Daniel Fandrich (3 Feb 2010)
57 - Changed the Watcom makefiles to make them easier to keep in sync with
58 Makefile.inc since that can't be included directly.
61 - Symbol CURL_FORMAT_OFF_T now obsoleted, will be removed in a future release,
62 symbol will not be available when building with CURL_NO_OLDIES defined. Use
63 of CURL_FORMAT_CURL_OFF_T is preferred since 7.19.0
65 Daniel Stenberg (1 Feb 2010)
66 - Using the multi_socket API, it turns out at times it seemed to "forget"
67 connections (which caused a hang). It turned out to be an existing (7.19.7)
68 bug in libcurl (that's been around for a long time) and it happened like
71 The app calls curl_multi_add_handle() to add a new easy handle, libcurl will
72 then set it to timeout in 1 millisecond so libcurl will tell the app about
75 The app's timeout fires off that there's a timeout, the app calls libcurl as
76 we so often document it:
79 res = curl_multi_socket_action(... TIMEOUT ...);
80 } while(CURLM_CALL_MULTI_PERFORM == res);
82 And this is the problem number one:
84 When curl_multi_socket_action() is called with no specific handle, but only
85 a timeout-action, it will *only* perform actions within libcurl that are
86 marked to run at this time. In this case, the request would go from INIT to
87 CONNECT and return CURLM_CALL_MULTI_PERFORM. When the app then calls libcurl
88 again, there's no timer set for this handle so it remains in the CONNECT
89 state. The CONNECT state is a transitional state in libcurl so it reports no
90 sockets there, and thus libcurl never tells the app anything more about that
91 easy handle/connection.
93 libcurl _does_ set a 1ms timeout for the handle at the end of
94 multi_runsingle() if it returns CURLM_CALL_MULTI_PERFORM, but since the loop
95 is instant the new job is not ready to run at that point (and there's no
96 code that makes libcurl call the app to update the timout for this new
97 timeout). It will simply rely on that some other timeout will trigger later
98 on or that something else will update the timeout callback. This makes the
99 bug fairly hard to repeat.
101 The fix made to adress this issue:
103 We introduce a loop in lib/multi.c around all calls to multi_runsingle() and
104 simply check for CURLM_CALL_MULTI_PERFORM internally. This has the added
105 benefit that this goes in line with my long-term wishes to get rid of the
106 CURLM_CALL_MULTI_PERFORM all together from the public API.
108 The downside of this fix, is that the counter we return in 'running_handles'
109 in several of our public functions then gets a slightly new and possibly
110 confusing behavior during times:
112 If an app adds a handle that fails to connect (very quickly) it may just
113 as well never appear as a 'running_handle' with this fix. Previously it
114 would first bump the counter only to get it decreased again at next call.
115 Even I have used that change in handle counter to signal "end of a
116 transfer". The only *good* way to find the end of a individual transfer
117 is calling curl_multi_info_read() to see if it returns one.
119 Of course, if the app previously did the looping before it checked the
120 counter, it really shouldn't be any new effect.
122 Yang Tse (26 Jan 2010)
123 - Constantine Sapuntzakis' and Joshua Kwan's work done in the last four months
124 relative to the asynchronous DNS lookups, along with with some integration
125 adjustments I have done are finally committed to CVS.
127 Currently these enhancements will benefit builds done using c-ares on any
128 platform as well as Windows builds using the default threaded resolver.
130 This release does not make generally available POSIX threaded DNS lookups
131 yet. There is no configure option to enable this feature yet. It is possible
132 to experimantally try this feature running configure with compiler flags that
133 make simultaneous definition of preprocessor symbols USE_THREADS_POSIX and
134 HAVE_PTHREAD_H, as well as whatever reentrancy compiler flags and linker ones
135 are required to link and properly use pthread_* functions on each platform.
137 Daniel Stenberg (26 Jan 2010)
138 - Mike Crowe made libcurl return CURLE_COULDNT_RESOLVE_PROXY when it is the
139 proxy that cannot be resolved when using c-ares. This matches the behaviour
140 when not using c-ares.
142 Björn Stenberg (23 Jan 2010)
143 - Added a new flag: -J/--remote-header-name. This option tells the
144 -O/--remote-name option to use the server-specified Content-Disposition
145 filename instead of extracting a filename from the URL.
147 Daniel Stenberg (21 Jan 2010)
148 - Chris Conroy brought support for RTSP transfers, and with it comes 8(!) new
149 libcurl options for controlling what to get and how to receive posssibly
150 interleaved RTP data.
152 Daniel Stenberg (20 Jan 2010)
153 - As was pointed out on the http-state mailing list, the order of cookies in a
154 HTTP Cookie: header _needs_ to be sorted on the path length in the cases
155 where two cookies using the same name are set more than once using
156 (overlapping) paths. Realizing this, identically named cookies must be
157 sorted correctly. But detecting only identically named cookies and take care
158 of them individually is harder than just to blindly and unconditionally sort
159 all cookies based on their path lengths. All major browsers also already do
160 this, so this makes our behavior one step closer to them in the cookie area.
162 Test case 8 was the only one that broke due to this change and I updated it
165 Daniel Stenberg (19 Jan 2010)
166 - David McCreedy brought a fix and a new test case (129) to make libcurl work
167 again when downloading files over FTP using ASCII and it turns out that the
168 final size of the file is not the same as the initial size the server
169 reported. This is very common since servers don't take the newline
170 conversions into account.
172 Kamil Dudka (14 Jan 2010)
173 - Suppressed side effect of OpenSSL configure checks, which prevented NSS from
174 being properly detected under certain circumstances. It had been caused by
175 strange behavior of pkg-config when handling PKG_CONFIG_LIBDIR. pkg-config
176 distinguishes among empty and non-existent environment variable in that case.
178 Daniel Stenberg (12 Jan 2010)
179 - Gil Weber reported a peculiar flaw with the multi interface when doing SFTP
180 transfers: curl_multi_fdset() would return -1 and not set and file
181 descriptors several times during a transfer of a single file. It turned out
182 to be due to two different flaws now fixed. Gil's excellent recipe helped me
185 Daniel Stenberg (11 Jan 2010)
186 - Made sure that the progress callback is repeatedly called at a regular
187 interval even during very slow connects.
189 - The tests/runtests.pl script now checks to see if the test case that runs is
190 present in the tests/data/Makefile.am and outputs a notice message on the
191 screen if not. Each test file has to be included in that Makefile.am to get
192 included in release archives and forgetting to add files there is a common
193 mistake. This is an attempt to make it harder to forget.
195 Daniel Stenberg (9 Jan 2010)
196 - Johan van Selst found and fixed a OpenSSL session ref count leak:
198 ossl_connect_step3() increments an SSL session handle reference counter on
199 each call. When sessions are re-used this reference counter may be
200 incremented many times, but it will be decremented only once when done (by
201 Curl_ossl_session_free()); and the internal OpenSSL data will not be freed
202 if this reference count remains positive. When a session is re-used the
203 reference counter should be corrected by explicitly calling
204 SSL_SESSION_free() after each consecutive SSL_get1_session() to avoid
205 introducing a memory leak.
207 (http://curl.haxx.se/bug/view.cgi?id=2926284)
209 Daniel Stenberg (7 Jan 2010)
210 - Make sure the progress callback is called repeatedly even during very slow
211 name resolves when c-ares is used for resolving.
213 Claes Jakobsson (6 Jan 2010)
214 - Julien Chaffraix fixed so that the fragment part in an URL is not sent
215 to the server anymore.
217 Kamil Dudka (3 Jan 2010)
218 - Julien Chaffraix eliminated a duplicated initialization in singlesocket().
220 Daniel Stenberg (2 Jan 2010)
221 - Make curl support --ssl and --ssl-reqd instead of the previous FTP-specific
222 versions --ftp-ssl and --ftp-ssl-reqd as these options are now used to
223 control SSL/TLS for IMAP, POP3 and SMTP as well in addition to FTP. The old
224 option names are still working but the new ones are the ones listed and
227 Daniel Stenberg (1 Jan 2010)
228 - Ingmar Runge enhanced libcurl's FTP engine to support the PRET command. This
229 command is a special "hack" used by the drftpd server, but even though it is
230 a custom extension I've deemed it fine to add to libcurl since this server
231 seems to survive and people keep using it and want libcurl to support
232 it. The new libcurl option is named CURLOPT_FTP_USE_PRET, and it is also
233 usable from the curl tool with --ftp-pret. Using this option on a server
234 that doesn't support this command will make libcurl fail.
236 I added test cases 1107 and 1108 to verify the functionality.
238 The PRET command is documented at
239 http://www.drftpd.org/index.php/Distributed_PASV
241 Yang Tse (30 Dec 2009)
242 - Steven M. Schweda improved VMS build system, and Craig A. Berry helped
243 with the patch and testing.
245 Daniel Stenberg (26 Dec 2009)
246 - Renato Botelho and Peter Pentchev brought a patch that makes the libcurl
247 headers work correctly even on FreeBSD systems before v8.
249 (http://curl.haxx.se/bug/view.cgi?id=2916915)
251 Daniel Stenberg (17 Dec 2009)
252 - David Byron fixed Curl_ossl_cleanup to actually call ENGINE_cleanup when
255 - Follow-up fix for the proxy fix I did for Jon Nelson's bug. It turned out I
256 was a bit too quick and broke test case 1101 with that change. The order of
257 some of the setups is sensitive. I now changed it slightly again to make
258 sure we do them in this order:
260 1 - parse URL and figure out what protocol is used in the URL
261 2 - prepend protocol:// to URL if missing
262 3 - parse name+password off URL, which needs to know what protocol is used
263 (since only some allows for name+password in the URL)
264 4 - figure out if a proxy should be used set by an option
265 5 - if no proxy option, check proxy environment variables
266 6 - run the protocol-specific setup function, which needs to have the proxy
269 Daniel Stenberg (15 Dec 2009)
270 - Jon Nelson found a regression that turned out to be a flaw in how libcurl
271 detects and uses proxies based on the environment variables. If the proxy
272 was given as an explicit option it worked, but due to the setup order
273 mistake proxies would not be used fine for a few protocols when picked up
274 from '[protocol]_proxy'. Obviously this broke after 7.19.4. I now also added
275 test case 1106 that verifies this functionality.
277 (http://curl.haxx.se/bug/view.cgi?id=2913886)
279 Daniel Stenberg (12 Dec 2009)
280 - IMAP, POP3 and SMTP support and their TLS versions (including IMAPS, POP3S
281 and SMTPS) are now supported. The current state may not yet be solid, but
282 the foundation is in place and the test suite has some initial support for
283 these protocols. Work will now persue to make them nice libcurl citizens
286 The work with supporting these new protocols was sponsored by
287 networking4all.com - thanks!
289 Daniel Stenberg (10 Dec 2009)
290 - Siegfried Gyuricsko found out that the curl manual said --retry would retry
291 on FTP errors in the transient 5xx range. Transient FTP errors are in the
292 4xx range. The code itself only tried on 5xx errors that occured _at login_.
293 Now the retry code retries on all FTP transfer failures that ended with a
296 (http://curl.haxx.se/bug/view.cgi?id=2911279)
298 - Constantine Sapuntzakis figured out a case which would lead to libcurl
299 accessing alredy freed memory and thus crash when using HTTPS (with
300 OpenSSL), multi interface and the CURLOPT_DEBUGFUNCTION and a certain order
301 of cleaning things up. I fixed it.
303 (http://curl.haxx.se/bug/view.cgi?id=2905220)
305 Daniel Stenberg (7 Dec 2009)
306 - Martin Storsjo made libcurl use the Expect: 100-continue header for posts
307 with unknown size. Previously it was only used for posts with a known size
308 larger than 1024 bytes.
310 Daniel Stenberg (1 Dec 2009)
311 - If the Expect: 100-continue header has been set by the application through
312 curl_easy_setopt with CURLOPT_HTTPHEADER, the library should set
313 data->state.expect100header accordingly - the current code (in 7.19.7 at
314 least) doesn't handle this properly. Martin Storsjo provided the fix!
316 Yang Tse (28 Nov 2009)
317 - Added Diffie-Hellman parameters to several test harness certificate files in
318 PEM format. Required by several stunnel versions used by our test harness.
320 Daniel Stenberg (28 Nov 2009)
321 - Markus Koetter provided a polished and updated version of Chad Monroe's TFTP
322 rework patch that now integrates TFTP properly into libcurl so that it can
323 be used non-blocking with the multi interface and more. BLKSIZE also works.
325 The --tftp-blksize option was added to allow setting the TFTP BLKSIZE from
328 Daniel Stenberg (26 Nov 2009)
329 - Extended and fixed the change I did on Dec 11 for the the progress
330 meter/callback during FTP command/response sequences. It turned out it was
331 really lame before and now the progress meter SHOULD get called at least
334 Daniel Stenberg (23 Nov 2009)
335 - Bjorn Augustsson reported a bug which made curl not report any problems even
336 though it failed to write a very small download to disk (done in a single
337 fwrite call). It turned out to be because fwrite() returned success, but
338 there was insufficient error-checking for the fclose() call which tricked
339 curl to believe things were fine.
341 Yang Tse (23 Nov 2009)
342 - David Byron modified Makefile.dist vc8 and vc9 targets in order to allow
343 finer granularity control when generating src and lib makefiles.
345 Yang Tse (22 Nov 2009)
346 - I modified configure to force removal of the curlbuild.h file included in
347 distribution tarballs for use by non-configure systems. As intended, this
348 would get overwriten when doing in-tree builds. But VPATH builds would end
349 having two curlbuild.h files, one in the source tree and another in the
350 build tree. With the modification I introduced 5 Nov 2009 this could become
351 an issue when running libcurl's test suite.
353 Daniel Stenberg (20 Nov 2009)
354 - Constantine Sapuntzakis identified a write after close, as the sockets were
355 closed by libcurl before the SSL lib were shutdown and they may write to its
356 socket. Detected to at least happen with OpenSSL builds.
358 - Jad Chamcham pointed out a bug with connection re-use. If a connection had
359 CURLOPT_HTTPPROXYTUNNEL enabled over a proxy, a subsequent request using the
360 same proxy with the tunnel option disabled would still wrongly re-use that
361 previous connection and the outcome would only be badness.
363 Yang Tse (18 Nov 2009)
364 - I modified the memory tracking system to make it intolerant with zero sized
365 malloc(), calloc() and realloc() function calls.
367 Daniel Stenberg (17 Nov 2009)
368 - Constantine Sapuntzakis provided another fix for the DNS cache that could
369 end up with entries that wouldn't time-out:
371 1. Set up a first web server that redirects (307) to a http://server:port
373 2. Have curl connect to the first web server using curl multi
375 After the curl_easy_cleanup call, there will be curl dns entries hanging
376 around with in_use != 0.
378 (http://curl.haxx.se/bug/view.cgi?id=2891591)
380 - Marc Kleine-Budde fixed: curl saved the LDFLAGS set during configure into
381 its pkg-config file. So -Wl stuff ended up in the .pc file, which is really
382 bad, and breaks if there are multiple -Wl in our LDFLAGS (which are in
383 PTXdist). bug #2893592 (http://curl.haxx.se/bug/view.cgi?id=2893592)
385 Kamil Dudka (15 Nov 2009)
386 - David Byron improved the configure script to use pkg-config to find OpenSSL
387 (and in particular the list of required libraries) even if a path is given
388 as argument to --with-ssl
390 Yang Tse (15 Nov 2009)
391 - I removed enable-thread / disable-thread configure option. These were only
392 placebo options. The library is always built as thread safe as possible on
395 Claes Jakobsson (14 Nov 2009)
396 - curl-config now accepts '--configure' to see what arguments was
397 passed to the configure script when building curl.
399 Daniel Stenberg (14 Nov 2009)
400 - Claes Jakobsson restored the configure functionality to detect NSS when
401 --with-nss is set but not "yes".
403 I think we can still improve that to check for pkg-config in that path etc,
404 but at least this patch brings back the same functionality we had before.
406 - Camille Moncelier added support for the file type SSL_FILETYPE_ENGINE for
407 the client certificate. It also disable the key name test as some engines
408 can select a private key/cert automatically (When there is only one key
409 and/or certificate on the hardware device used by the engine)
411 Yang Tse (14 Nov 2009)
412 - Constantine Sapuntzakis provided the fix that ensures that an SSL connection
413 won't be reused unless protection level for peer and host verification match.
415 I refactored how preprocessor symbol _THREAD_SAFE definition is done.
417 Kamil Dudka (12 Nov 2009)
418 - Kevin Baughman provided a fix preventing libcurl-NSS from crash on doubly
419 closed NSPR descriptor. The issue was hard to find, reported several times
420 before and always closed unresolved. More info at the RH bug:
421 https://bugzilla.redhat.com/534176
423 - libcurl-NSS now tries to reconnect with TLS disabled in case it detects
424 a broken TLS server. However it does not happen if SSL version is selected
425 manually. The approach was originally taken from PSM. Kaspar Brand helped me
426 to complete the patch. Original bug reports:
427 https://bugzilla.redhat.com/525496
428 https://bugzilla.redhat.com/527771
430 Yang Tse (12 Nov 2009)
431 - I modified configure script to make the getaddrinfo function check also
432 verify if the function is thread safe.
434 Yang Tse (11 Nov 2009)
435 - Marco Maggi reported that compilation failed when configured --with-gssapi
436 and GNU GSS installed due to a missing mutual exclusion of header files in
437 the Kerberos 5 code path. He also verified that my patch worked for him.
439 Daniel Stenberg (11 Nov 2009)
440 - Constantine Sapuntzakis posted bug #2891595
441 (http://curl.haxx.se/bug/view.cgi?id=2891595) which identified how an entry
442 in the DNS cache would linger too long if the request that added it was in
443 use that long. He also provided the patch that now makes libcurl capable of
444 still doing a request while the DNS hash entry may get timed out.
446 - Christian Schmitz noticed that the progress meter/callback was not properly
447 used during the FTP connection phase (after the actual TCP connect), while
448 it of course should be. I also made the speed check get called correctly so
449 that really slow servers will trigger that properly too.
451 Kamil Dudka (5 Nov 2009)
452 - Dropped misleading timeouts in libcurl-NSS and made sure the SSL socket works
453 in non-blocking mode.
455 Yang Tse (5 Nov 2009)
456 - I removed leading 'curl' path on the 'curlbuild.h' include statement in
457 curl.h, adjusting auto-makefiles include path, to enhance portability to
458 OS's without an orthogonal directory tree structure such as OS/400.
460 Daniel Stenberg (4 Nov 2009)
461 - I fixed several problems with the transfer progress meter. It showed the
462 wrong percentage for small files, most notable for <1000 bytes and could
463 easily end up showing more than 100% at the end. It also didn't show any
464 percentage, transfer size or estimated transfer times when transferring
467 Version 7.19.7 (4 November 2009)
469 Daniel Stenberg (2 Nov 2009)
470 - As reported independent by both Stan van de Burgt and Didier Brisebourg,
471 CURLINFO_SIZE_DOWNLOAD (the -w variable size_download) didn't work when
472 getting data from ldap!
474 Daniel Stenberg (31 Oct 2009)
475 - Gabriel Kuri reported a problem with CURLINFO_CONTENT_LENGTH_DOWNLOAD if the
476 download was 0 bytes, as libcurl would then return the size as unknown (-1)
477 and not 0. I wrote a fix and test case 566 to verify it.
479 Daniel Stenberg (30 Oct 2009)
480 - Liza Alenchery mentioned a problem with re-used SCP connection when a bad
481 auth is used, as it caused a crash. I failed to repeat the issue, but still
482 made a change that now forces the TCP connection used for a freed SCP
483 session to get closed and not be re-used.
485 - "Tom" posted a bug report that mentioned how libcurl did wrong when doing a
486 POST using a read callback, with Digest authentication and
487 "Transfer-Encoding: chunked" enforced. I would then cause the first request
488 to be wrongly sent and then basically hang until the server closed the
489 connection. I fixed the problem and added test case 565 to verify it.
491 Daniel Stenberg (25 Oct 2009)
492 - Dima Barsky made the curl cookie parser accept cookies even with blank or
493 unparsable expiry dates and then treat them as session cookies - previously
494 libcurl would reject cookies with a date format it couldn't parse. Research
495 shows that the major browser treat such cookies as session cookies. I
496 modified test 8 and 31 to verify this.
498 Daniel Stenberg (21 Oct 2009)
499 - Attempt to use pkg-config for finding out libssh2 installation details
502 - A patch in bug report #2883177 (http://curl.haxx.se/bug/view.cgi?id=2883177)
503 by Johan van Selst introduced the --crlfile option to curl, which makes curl
504 tell libcurl about a file with CRL (certificate revocation list) data to
507 Daniel Stenberg (18 Oct 2009)
508 - Ray Dassen provided a patch in Debian's bug tracker (bug number #551461)
509 that now makes curl_getdate(3) actually handles RFC 822 formatted dates that
510 use the "single letter military timezones".
511 http://www.rfc-ref.org/RFC-TEXTS/822/chapter5.html has the details.
513 - Fixed memory leak in the SCP/SFTP code as it never freed the knownhosts
516 - John Dennis filed bug report #2873666
517 (http://curl.haxx.se/bug/view.cgi?id=2873666) which identified a problem
518 which made libcurl loop infinitely when given incorrect credentials when
519 using HTTP GSS negotiate authentication. He also provided a small and simple
522 - Kevin Baughman found a double close() problem with libcurl-NSS, as when
523 libcurl called NSS to close the SSL "session" it also closed the actual
526 Yang Tse (17 Oct 2009)
527 - Bug report #2866724 indicated
528 (http://curl.haxx.se/bug/view.cgi?id=2866724) that curl on Windows failed
529 when writing files whose file names originally contained characters which
530 are not valid for file names on Windows. Dan Fandrich provided an initial
531 patch and another revised one to fix this issue.
533 Daniel Stenberg (1 Oct 2009)
534 - Tom Mueller correctly reported in bug report #2870221
535 (http://curl.haxx.se/bug/view.cgi?id=2870221) that libcurl returned an
536 incorrect return code from the internal trynextip() function which caused
537 him grief. This is a regression that was introduced in 7.19.1 and I find it
538 strange it hasn't hit us harder, but I won't persue into figuring out
541 - Constantine Sapuntzakis: The current implementation will always set
542 SO_SNDBUF to CURL_WRITE_SIZE even if the SO_SNDBUF starts out larger. The
543 patch doesn't do a setsockopt if SO_SNDBUF is already greater than
544 CURL_WRITE_SIZE. This should help folks who have set up their computer with
547 Daniel Stenberg (27 Sep 2009)
548 - I introduced a maximum limit for received HTTP headers. It is controlled by
549 the define CURL_MAX_HTTP_HEADER which is even exposed in the public header
550 file to allow for users to fairly easy rebuild libcurl with a modified
551 limit. The rationale for a fixed limit is that libcurl is realloc()ing a
552 buffer to be able to put a full header into it, so that it can call the
553 header callback with the entire header, but that also risk getting it into
554 trouble if a server by mistake or willingly sends a header that is more or
555 less without an end. The limit is set to 100K.
557 Daniel Stenberg (26 Sep 2009)
558 - John P. McCaskey posted a bug report that showed how libcurl did wrong when
559 saving received cookies with no given path, if the path in the request had a
560 query part. That is means a question mark (?) and characters on the right
561 side of that. I wrote test case 1105 and fixed this problem.
563 Kamil Dudka (26 Sep 2009)
564 - Implemented a protocol independent way to specify blocking direction, used by
565 transfer.c for blocking. It is currently used only by SCP and SFTP protocols.
566 This enhancement resolves an issue with 100% CPU usage during SFTP upload,
569 Daniel Stenberg (25 Sep 2009)
570 - Chris Mumford filed bug report #2861587
571 (http://curl.haxx.se/bug/view.cgi?id=2861587) identifying that libcurl used
572 the OpenSSL function X509_load_crl_file() wrongly and failed if it would
573 load a CRL file with more than one certificate within. This is now fixed.
575 Daniel Stenberg (16 Sep 2009)
576 - Sven Anders reported that we introduced a cert verfication flaw for OpenSSL-
577 powered libcurl in 7.19.6. If there was a X509v3 Subject Alternative Name
578 field in the certficate it had to match and so even if non-DNS and non-IP
579 entry was present it caused the verification to fail.
581 Daniel Fandrich (15 Sep 2009)
582 - Moved the libssh2 checks after the SSL library checks. This helps when
583 statically linking since libssh2 needs the SSL library link flags to be
584 set up already to satisfy its dependencies. This wouldn't be necessary if
585 the libssh2 configure check was changed to use pkg-config since the
586 --static flag would add the dependencies automatically.
588 Yang Tse (14 Sep 2009)
589 - Revert Joshua Kwan's patch committed 11 Sep 2009.
591 Some systems poll function sets POLLHUP in revents without setting
592 POLLIN, and sets POLLERR without setting POLLIN and POLLOUT. In some
593 libcurl code execution paths this could trigger busy wait loops with
594 high CPU usage until a timeout condition aborted the loop.
596 The reverted patch addressed the above issue for a very specific case,
597 when awaiting c-ares to resolve. A libcurl-wide fix for Curl_poll now
600 Guenter Knauf (11 Sep 2009)
601 - Joshua Kwan provided a patch to pass POLLERR / POLLHUP back to c-ares.
602 This fixes a loop problem with high CPU usage.
604 Daniel Stenberg (10 Sep 2009)
605 - Claes Jakobsson fixed a problem with cookie expiry dates at exctly the epoch
606 start second "Thu Jan 1 00:00:00 GMT 1970" as the date parser then returns 0
607 which internally then is treated as a session cookie. That particular date
608 is now made to get the value of 1.
610 Daniel Stenberg (2 Sep 2009)
611 - Daniel Johnson found a flaw in the code converting sftp-errors to libcurl
614 Daniel Stenberg (1 Sep 2009)
615 - Peter Sylvester made a debug feature for Curl_resolv() that now will force
616 libcurl to resolve 'localhost' whatever name you use in the URL *if* you set
617 the --interface option to (exactly) "LocalHost". This will enable us to
618 write tests for custom hosts names but still use a local host server.
620 - configure now tries to use pkg-config for a number of sub-dependencies even
621 when cross-compiling. The key to success is then you properly setup
622 PKG_CONFIG_PATH before invoking configure.
624 I also improved how NSS is detected by trying nss-config if pkg-config isn't
625 present, and as a last resort just use the lib name and force the user to
626 setup the LIBS/LDFLAGS/CFLAGS etc properly. The previous last resort would
627 add a range of various libs that would almost never be quite correct.
629 Daniel Stenberg (31 Aug 2009)
630 - When using the multi interface with FTP and you asked for NOBODY, you did no
631 QUOTE commands and the request used the same path as the connection had
632 already changed to, it would decide that no commands would be necessary for
633 the "DO" action and that was not handled properly but libcurl would instead
636 Kamil Dudka (28 Aug 2009)
637 - Improved error message for not matching certificate subject name in
638 libcurl-NSS. Originally reported at:
639 https://bugzilla.redhat.com/show_bug.cgi?id=516056#c9
641 Patrick Monnerat (24 Aug 2009)
642 - Introduced a SYST-based test to properly set-up name format when dealing
643 with the OS/400 FTP server.
645 - Fixed an ftp_readresp() bug preventing detection of failing control socket
646 and causing FTP client to loop forever.
648 Daniel Stenberg (24 Aug 2009)
649 - Marc de Bruin pointed out that configure --with-gnutls=PATH didn't work
650 properly and provided a fix. http://curl.haxx.se/bug/view.cgi?id=2843008
652 - Eric Wong introduced support for the new option -T. (dot) that makes curl
653 read stdin in a non-blocking fashion. This also brings back -T- (minus) to
654 the previous blocking behavior since it could break stuff for people at
657 Michal Marek (21 Aug 2009)
658 - With CURLOPT_PROXY_TRANSFER_MODE, avoid sending invalid URLs like
659 ftp://example.com;type=i if the user specified ftp://example.com without the
662 Daniel Stenberg (21 Aug 2009)
663 - Andre Guibert de Bruet pointed out a missing return code check for a
664 strdup() that could lead to segfault if it returned NULL. I extended his
665 suggest patch to now have Curl_retry_request() return a regular return code
666 and better check that.
668 - Lots of good work by Krister Johansen, mostly related to pipelining:
670 Fix SIGSEGV on free'd easy_conn when pipe unexpectedly breaks
671 Fix data corruption issue with re-connected transfers
672 Fix use after free if we're completed but easy_conn not NULL
674 Kamil Dudka (13 Aug 2009)
675 - Changed NSS code to not ignore the value of ssl.verifyhost and produce more
676 verbose error messages. Originally reported at:
677 https://bugzilla.redhat.com/show_bug.cgi?id=516056
679 Daniel Stenberg (12 Aug 2009)
680 - Karl Moerder fixed the Makefile.vc* makefiles to include the new file
681 nonblock.c so that they work fine again
683 - I expanded test 517 with a bunch of more dates that originate from the
684 Chrome browser test suite. It turns out most of them get parsed the same
687 Version 7.19.6 (12 August 2009)
689 Daniel Stenberg (12 Aug 2009)
690 - Carsten Lange reported a bug and provided a patch for TFTP upload and the
691 sending of the TSIZE option. I don't like fixing bugs just hours before
692 a release, but since it was broken and the patch fixes this for him I decided
695 Daniel Stenberg (11 Aug 2009)
696 - Peter Sylvester made the HTTPS test server use specific certificates for
697 each test, so that the test suite can now be used to actually test the
698 verification of cert names etc. This made an error show up in the OpenSSL-
699 specific code where it would attempt to match the CN field even if a
700 subjectAltName exists that doesn't match. This is now fixed and verified
703 - Benbuck Nason posted the bug report #2835196
704 (http://curl.haxx.se/bug/view.cgi?id=2835196), fixing a few compiler
705 warnings when mixing ints and bools.
707 Daniel Fandrich (10 Aug 2009)
708 - Fixed a memory leak in the FTP code and an off-by-one heap buffer overflow.
710 Daniel Fandrich (9 Aug 2009)
711 - Fixed some memory leaks in the command-line tool that caused most of the
712 torture tests to fail.
714 Daniel Stenberg (2 Aug 2009)
715 - Curt Bogmine reported a problem with SNI enabled on a particular server. We
716 should introduce an option to disable SNI, but as we're in feature freeze
717 now I've addressed the obvious bug here (pointed out by Peter Sylvester): we
718 shouldn't try to enable SNI when SSLv2 or SSLv3 is explicitly selected.
719 Code for OpenSSL and GnuTLS was fixed. NSS doesn't seem to have a particular
720 option for SNI, or are we simply not using it?
722 Daniel Stenberg (1 Aug 2009)
723 - Scott Cantor posted the bug report #2829955
724 (http://curl.haxx.se/bug/view.cgi?id=2829955) mentioning the recent SSL cert
725 verification flaw found and exploited by Moxie Marlinspike. The presentation
726 he did at Black Hat is available here:
727 https://www.blackhat.com/html/bh-usa-09/bh-usa-09-archives.html#Marlinspike
729 Apparently at least one CA allowed a subjectAltName or CN that contain a
730 zero byte, and thus clients that assumed they would never have zero bytes
731 were exploited to OK a certificate that didn't actually match the site. Like
732 if the name in the cert was "example.com\0theatualsite.com", libcurl would
733 happily verify that cert for example.com.
735 libcurl now better uses the length of the extracted name, not using the zero
736 termination for getting the string length.
738 This fixing only made and needed in OpenSSL interfacing code.
740 - Tanguy Fautre pointed out that OpenSSL's function RAND_screen() (present
741 only in some OpenSSL installs - like on Windows) isn't thread-safe and we
742 agreed that moving it to the global_init() function is a decent way to deal
745 - Alexander Beedie provided the patch for a noproxy problem: If I have set
746 CURLOPT_NOPROXY to "*", or to a host that should not use a proxy, I actually
747 could still end up using a proxy if a proxy environment variable was set.
749 Daniel Stenberg (27 Jul 2009)
750 - All the quote options (CURLOPT_QUOTE, CURLOPT_POSTQUOTE and
751 CURLOPT_PREQUOTE) now accept a preceeding asterisk before the command to
752 send when using FTP, as a sign that libcurl shall simply ignore the response
753 from the server instead of treating it as an error. Not treating a 400+ FTP
754 response code as an error means that failed commands will not abort the
755 chain of commands, nor will they cause the connection to get disconnected.
757 Daniel Stenberg (26 Jul 2009)
758 - Johan van Selst posted bug report #2825989
759 (http://curl.haxx.se/bug/view.cgi?id=2825989) pointing out that
760 OpenSSL-powered libcurl didn't support the SHA-2 digest algorithm, and
761 provided the solution too: to use OpenSSL_add_all_algorithms() in addition
762 to the older SSLeay_* alternative. OpenSSL_add_all_algorithms was added in
765 Daniel Stenberg (23 Jul 2009)
766 - Added CURLOPT_SSH_KNOWNHOSTS, CURLOPT_SSH_KEYFUNCTION, CURLOPT_SSH_KEYDATA.
767 They introduce known_host support for SSH keys to libcurl. See docs for
768 details. Note that this feature depends on a new enough libssh2 version, to
769 be supported in libssh2 1.2 and later (or current git repo at this time).
771 Michal Marek (22 Jul 2009)
772 - David Binderman found a memory and fd leak in lib/gtls.c:load_file()
773 (https://bugzilla.novell.com/523919). When looking at the code, I found that
774 also the ptr pointer can leak.
776 Kamil Dudka (20 Jul 2009)
777 - Claes Jakobsson improved the support for client certificates handling in
778 NSS-powered libcurl. Now the client certificates can be selected
779 automatically by a NSS built-in hook. Additionally pre-login to all PKCS11
780 slots is no more performed. It used to cause problems with HW tokens.
782 - Fixed reference counting for NSS client certificates. Now the PEM reader
783 module should be always properly unloaded on Curl_nss_cleanup(). If the
784 unload fails though, libcurl will try to reuse the already loaded instance.
786 Daniel Fandrich (15 Jul 2009)
787 - Added nonblock.c to the non-automake makefiles (note that the dependencies
788 in the Watcom makefiles aren't quite correct).
790 Michal Marek (15 Jul 2009)
791 - Changed the description of CURLINFO_OS_ERRNO to make it clear that the
792 errno is not reset on success.
794 Guenter Knauf (14 Jul 2009)
795 - renamed generated config.h to curl_config.h to avoid any future clashes
796 with config.h from other projects.
798 Daniel Stenberg (9 Jul 2009)
799 - Eric Wong introduced curlx_nonblock() that the curl tool now (re-)uses for
800 setting a file descriptor non-blocking. Used by the functionality Eric
801 himself brough on June 15th.
803 Daniel Stenberg (8 Jul 2009)
804 - Constantine Sapuntzakis posted bug report #2813123
805 (http://curl.haxx.se/bug/view.cgi?id=2813123) and an a patch that fixes the
808 Url A is accessed using auth. Url A redirects to Url B (on a different
809 server0. Url B reuses a persistent connection. Url B has auth, even though
810 it's on a different server.
812 Note: if Url B does not reuse a persistent connection, auth is not sent.
816 data->state.first_host is not initialized becuase Curl_http_connect is not
817 called when a connection is reused.
821 move initialization of data->state.first_host to Curl_http. No code before
822 Curl_http uses data->state.first_host anyway.
824 Guenter Knauf (4 Jul 2009)
825 - Markus Koetter provided a patch to avoid getnameinfo() usage which broke a
826 couple of both IPv4 and IPv6 autobuilds.
828 Daniel Stenberg (29 Jun 2009)
829 - Markus Koetter made CURLOPT_FTPPORT (and curl's -P/--ftpport) support a port
830 range if given colon-separated after the host name/address part. Like
831 "192.168.0.1:2000-10000"
833 - Modified the separators used for CURLOPT_CERTINFO in multi-part outputs. I
834 don't know how they got wrong in the first place, but using this output
835 format makes it possible to quite easily separate the string into an array
838 Daniel Fandrich (16 June 2009)
839 - Added a few more compiler warning options for gcc.
841 Daniel Stenberg (16 Jun 2009)
842 - Reuven Wachtfogel made curl -o - properly produce a binary output on windows
843 (no newline translations). Use -B/--use-ascii if you rather get the ascii
846 Michal Marek (16 Jun 2009)
847 - When doing non-anonymous ftp via http proxies and the password is not
848 provided in the url, add it there (squid needs this).
850 Daniel Stenberg (15 Jun 2009)
853 This allows curl(1) to be used as a client-side tunnel for arbitrary stream
854 protocols by abusing chunked transfer encoding in both the HTTP request and
855 HTTP response. This requires server support for sending a response while a
856 request is still being read, of course.
858 If attempting to read from stdin returns EAGAIN, then we pause our sender.
859 This leaves curl to attempt to read from the socket while reading from stdin
860 (and thus sending) is paused.
862 This change was needed to allow successfully tunneling the git protocol over
863 HTTP (--no-buffer is needed, as well).
865 Patrick Monnerat (15 Jun 2009)
866 - Replaced use of standard C library rand()/srand() by our own pseudo-random
869 Yang Tse (11 Jun 2009)
870 - I adapted testcurl script to allow building test harness programs when
871 cross-compiling for a *-*-mingw* host.
873 Daniel Stenberg (10 Jun 2009)
874 - Fabian Keil ran clang on the (lib)curl code, found a bunch of warnings and
875 contributed a range of patches to fix them.
877 Yang Tse (10 Jun 2009)
878 - I introduced configure script option --enable-curldebug which now allows
879 the decoupled enabling or disabling of the curl debug memory tracking
880 feature from the --enable-debug option which no longer controls this.
882 curl --version will list 'Debug' feature for debug enabled builds, and
883 will list 'TrackMemory' feature for curl debug memory tracking capable
884 builds. These features are independent and can be controlled when running
885 the configure script. When --enable-debug is given both features will be
886 enabled, unless some restriction prevents memory tracking from being used.
888 Internally, definition of preprocessor symbol DEBUGBUILD restricts code
889 which is only compiled for debug enabled builds. And symbol CURLDEBUG is
890 used to differentiate code which is _only_ used for memory tracking.
892 Yang Tse (9 Jun 2009)
893 - Daniel Steinberg pointed out that Curl_FormInit() in formdata.c was not
894 initializing the fread callback pointer and this triggered a compiler
895 warning, also provided a friendly suggestion on how to fix it.
897 Daniel Stenberg (8 Jun 2009)
898 - Claes Jakobsson provided a patch for libcurl-NSS that fixed a bad refcount
899 issue with client certs that caused issues like segfaults.
900 http://curl.haxx.se/mail/lib-2009-05/0316.html
902 - Triggered by bug report #2798852 and the patch in there, I fixed configure
903 to detect gnutls build options with pkg-config only and not libgnutls-config
904 anymore since GnuTLS has stopped distributing that tool. If an explicit path
905 is given to configure, we will instead guess on how to link and use that
906 lib. I did not use the patch from the bug report.
908 Yang Tse (8 Jun 2009)
909 - Igor Novoseltsev adjusted Makefile.vxworks to get sources and headers
910 included from Makefile.inc, and provided docs\INSTALL VxWorks section.
912 - I removed buildconf.bat from release and daily snapshot archives. This
913 file is only for CVS tree checkout builds.
915 Daniel Stenberg (8 Jun 2009)
916 - Eric Wong fixed --no-buffer to actually switch off output buffering. Been
919 Bill Hoffman (6 Jun 2009)
920 - Added some cmake docs and fixed socklen_t in the build.
922 Yang Tse (5 Jun 2009)
923 - John E. Malmberg provided VMS specific patch: "This fixes an existing bug
924 in urlglob.c where it was not converting the Curl Unix exit code to a VMS
925 DCL compatible exit code. This fix required the enhancement described next.
926 This also adds an enhancement to main.c so that when curl is run under a
927 Unix shell like Bash on VMS, it will return the standard Unix exit codes
928 and messages." And another patch for docs/examples.
930 I introduced os-specific.c and os-specific.h for use in curl tool code
931 and adjusted John E. Malmberg's patch placement to use these new files
932 as an effort to prevent main.c from growing ad infinitum. Code already
933 existing in main.c which is OS specific should be moved into these files.
935 Daniel Stenberg (4 June 2009)
936 - Setting the Content-Length: header from your app when you do a POST or PUT
937 is almost always a VERY BAD IDEA. Yet there are still apps out there doing
938 this, and now recently it triggered a bug/side-effect in libcurl as when
939 libcurl sends a POST or PUT with NTLM, it sends an empty post first when it
940 knows it will just get a 401/407 back. If the app then replaced the
941 Content-Length header, it caused the server to wait for input that libcurl
942 wouldn't send. Aaron Oneal reported this problem in bug report #2799008
943 (http://curl.haxx.se/bug/view.cgi?id=2799008) and helped us verify the fix.
945 Yang Tse (4 Jun 2009)
946 - Igor Novoseltsev provided patches and information, that after some
947 adjustments to better fit curl's way of doing things, have resulted
948 in the posibility of building libcurl for VxWorks.
950 Daniel Fandrich (2 June 2009)
951 - Checked in a Google Android make file. To use it, you must first
952 create a config.h file by running configure in the Android environment,
953 which doesn't seem to be easy to do. If no easy way can be found, a
954 static config-android.h may need to be created and checked in to the
957 Daniel Stenberg (1 June 2009)
958 - Claes Jakobsson fixed the configure script to better find and use NSS
961 Yang Tse (1 Jun 2009)
962 - John E. Malmberg provided a VMS specific clean-up for curl.h, and pointed
963 out that the configure script was failing to detect the timeval struct on
964 VMS when building with _XOPEN_SOURCE_EXTENDED undefined due to definition
965 taking place in socket.h instead of time.h. I have adjusted configure
966 script to also include this header when checking struct timeval.
968 Daniel Stenberg (27 May 2009)
969 - Frank McGeough provided a small OpenSSL #include fix to make libcurl compile
970 fine with Nokia 5th edition 1.0 SDK for Symbian.
972 - Andre Guibert de Bruet found a call to a OpenSSL function that didn't check
973 for a failure properly.
975 - Mike Crowe pointed out that setting CURLOPT_USERPWD to NULL used to clear
976 the auth credentials back in 7.19.0 and earlier while now you have to set ""
977 to get the same effect. His patch brings back the ability to use NULL.
979 - Claes Jakobsson fixed libcurl-NSS to build fine even without the
980 PK11_CreateGenericObject() function.
982 Daniel Stenberg (25 May 2009)
983 - bug report #2796358 (http://curl.haxx.se/bug/view.cgi?id=2796358) pointed
984 out that the cookie parser would leak memory when it parses cookies that are
985 received with domain, path etc set multiple times in the same header. While
986 such a cookie is questionable, they occur in the wild and libcurl no longer
987 leaks memory for them. I added such a header to test case 8.
989 Daniel Fandrich (22 May 2009)
990 - Removed some obsolete digest code that caused a valgrind error in test 551.
992 Daniel Fandrich (20 May 2009)
993 - Added "non-existing host" test keywords to make it easy to skip those
994 tests on machines that have broken DNS configurations (such as
995 those configured to use OpenDNS).
997 Daniel Stenberg (19 May 2009)
998 - Kamil Dudka brought the patch from the Redhat bug entry
999 https://bugzilla.redhat.com/show_bug.cgi?id=427966 which was libcurl closing
1000 a bad file descriptor when closing down the FTP data connection. Caolan
1001 McNamara seems to be the original author of it.
1003 Version 7.19.5 (18 May 2009)
1005 Daniel Stenberg (17 May 2009)
1006 - James Bursa posted a patch to the mailing list that fixed a problem with
1007 no_proxy which made it not skip the proxy if the URL entered contained a
1008 user name. I added test case 1101 to verify.
1010 Daniel Stenberg (11 May 2009)
1011 - Balint Szilakszi reported a memory leak when libcurl did gzip decompression
1012 of streams that had some parts (legitimately) missing. We now provide and use
1013 a proper cleanup function for the content encoding submodule.
1014 http://curl.haxx.se/mail/lib-2009-05/0092.html
1016 - Kamil Dudka provided a fix for libcurl-NSS reported by Michael Cronenworth
1017 at https://bugzilla.redhat.com/show_bug.cgi?id=453612#c12
1019 If an incorrect password is given while loading a private key, libcurl ends
1020 up in an infinite loop consuming memory. The bug is critical.
1022 - I fixed the problem with doing NTLM, POST and then following a 302 redirect,
1023 as reported by Ebenezer Ikonne (on curl-users) and Laurent Rabret (on
1024 curl-library). The transfer was mistakenly marked to get more data to send
1025 but since it didn't actually have that, it just hung there...
1027 Daniel Stenberg (10 May 2009)
1028 - Andre Guibert de Bruet correctly pointed out an over-alloc with one wasted
1029 byte in the digest code.
1031 Yang Tse (9 May 2009)
1032 - Removed DOS and TPF package's subdirectory Makefile.am, it was only used
1033 to include some files in the distribution tarball serving no other purpose.
1034 Files from the DOS and TPF subdirectories are now included in the EXTRA_DIST
1035 of the Makefile in the parent subdirectory.
1037 Yang Tse (8 May 2009)
1038 - Changed host name literal in several tests to one under the haxx.se domain.
1040 - Renamed vc6 workspace and project files to avoid filename clash when used
1041 for conversion to later VS versions.
1043 Daniel Stenberg (8 May 2009)
1044 - Constantine Sapuntzakis fixed bug report #2784055
1045 (http://curl.haxx.se/bug/view.cgi?id=2784055) identifying a problem to
1046 connect to SOCKS proxies when using the multi interface. It turned out to
1047 almost not work at all previously. We need to wait for the TCP connect to
1048 be properly verified before doing the SOCKS magic.
1050 There's still a flaw in the FTP code for this.
1052 Daniel Stenberg (7 May 2009)
1053 - Made the SO_SNDBUF setting for the data connection socket for ftp uploads as
1054 well. See change 28 Apr 2009.
1056 Yang Tse (7 May 2009)
1057 - Fixed an issue affecting FTP transfers, introduced with the transfer.c
1058 patch committed May 4.
1060 Daniel Stenberg (7 May 2009)
1061 - Man page *roff problems fixed thanks to input from Colin Watson. Problems
1062 reported in the Debian package.
1064 - Vijay G filed bug report #2723236
1065 (http://curl.haxx.se/bug/view.cgi?id=2723236) identifying a problem with
1066 libcurl's TFTP code and its lack of dealing with the OACK packet.
1068 Yang Tse (5 May 2009)
1069 - Fixed the --ftp-port address of test #251 to the CLIENTIP address, and
1070 reverted the change affecting test suite harness committed 4 May.
1072 Daniel Stenberg (5 May 2009)
1073 - Inspired by Michael Smith's session id fix for OpenSSL, I did the
1074 corresponding fix in the GnuTLS code: make sure to store the new session id
1075 in case the previous re-used one is rejected.
1077 Daniel Stenberg (4 May 2009)
1078 - Michael Smith posted bug report #2786255
1079 (http://curl.haxx.se/bug/view.cgi?id=2786255) with a patch, identifying how
1080 libcurl did not deal with SSL session ids properly if the server rejected a
1081 re-use of one. Starting now, it will forget the rejected one and remember
1082 the new. This change was for OpenSSL only, it is likely that other SSL lib
1083 code needs similar fixes.
1085 Yang Tse (4 May 2009)
1086 - Applied David McCreedy's "transfer.c fixes for CURL_DO_LINEEND_CONV and
1087 non-ASCII platform HTTP requests" patch addressing two HTTP PUT problems:
1088 1) On non-ASCII platforms not all of the protocol portions of the PUT are
1089 being translated to ASCII. 2) On all platforms the line endings of part of
1090 the protocol portions are mangled from CRLF to CRCRLF if data->set.crlf or
1091 data->set.prefer_ascii are set (depending on CURL_DO_LINEEND_CONV).
1093 - Applied David McCreedy's patch to fix test suite harness to allow test FTP
1094 server and client on different machines, providing FTP client address when
1095 running the FTP test server.
1097 Daniel Fandrich (3 May 2009)
1098 - Added and disabled test case 563 which shows KNOWN_BUGS #59. The bug
1099 report failed to mention that a proxy must be used to reproduce it.
1101 Yang Tse (2 May 2009)
1102 - Use a build-time configured curl_socklen_t data type instead of socklen_t.
1104 Yang Tse (1 May 2009)
1105 - Applied David McCreedy's patches "TPF-platform specific changes to various
1106 files" and "http.c fix to Curl_proxyCONNECT for non-ASCII platforms", the
1107 former with minor edits.
1109 Daniel Stenberg (30 Apr 2009)
1110 - I was going to fix issue #59 in KNOWN_BUGS
1112 If the CURLOPT_PORT option is used on an FTP URL like
1113 "ftp://example.com/file;type=A" the ";type=A" is stripped off.
1115 I added test case 562 to verify, only to find out that I couldn't repeat
1116 this bug so I hereby consider it not a bug anymore!
1118 Daniel Stenberg (29 Apr 2009)
1119 - Based on bug report #2723219 (http://curl.haxx.se/bug/view.cgi?id=2723219)
1120 I've now made TFTP "connections" not being kept for re-use within libcurl.
1121 TFTP is UDP-based so the benefit was really low (if even existing) to begin
1122 with so instead of tracking down to fix this problem we instead removed the
1123 re-use. I also enabled test case 1099 that I wrote a few days ago to verify
1124 that this change fixes the reported problem.
1126 Daniel Stenberg (28 Apr 2009)
1127 - Constantine Sapuntzakis filed bug report #2783090
1128 (http://curl.haxx.se/bug/view.cgi?id=2783090) pointing out that on windows
1129 we need to grow the SO_SNDBUF buffer somewhat to get really good upload
1130 speeds. http://support.microsoft.com/kb/823764 has the details. Friends
1131 confirmed that simply adding 32 to CURL_MAX_WRITE_SIZE is enough.
1133 - Bug report #2709004 (http://curl.haxx.se/bug/view.cgi?id=2709004) by Tim
1134 Chen pointed out how curl couldn't upload with resume when reading from a
1137 This ended up with the introduction of a new return code for the
1138 CURLOPT_SEEKFUNCTION callback that basically says that the seek failed but
1139 that libcurl may try to resolve the situation anyway. In our case this means
1140 libcurl will attempt to instead read that much data from the stream instead
1141 of seeking and that way curl can now upload with resume when data is read
1144 Daniel Stenberg (26 Apr 2009)
1145 - Bug report #2779733 (http://curl.haxx.se/bug/view.cgi?id=2779733) by Sven
1146 Wegener pointed out that CURLINFO_APPCONNECT_TIME didn't work with the multi
1147 interface and provided a patch that fixed the problem!
1149 Daniel Stenberg (24 Apr 2009)
1150 - Kamil Dudka fixed another NSS-related leak when client certs were used.
1152 - Bug report #2779245 (http://curl.haxx.se/bug/view.cgi?id=2779245) by Rainer
1153 Koenig pointed out that the man page didn't tell that the *_proxy
1154 environment variables can be specified lower case or UPPER CASE and the
1155 lower case takes precedence,
1157 Daniel Fandrich (21 Apr 2009)
1158 - Added new libcurl source files to Amiga, RiscOS and VC6 build files.
1160 Yang Tse (21 Apr 2009)
1161 - Moved potential inclusion of system's malloc.h and memory.h header files to
1162 setup_once.h. Inclusion of each header file is based on the definition of
1163 NEED_MALLOC_H and NEED_MEMORY_H respectively.
1165 Renamed libcurl's memory.h to curl_memory.h
1167 Daniel Stenberg (20 Apr 2009)
1168 - Leanic Lefever reported a crash and did some detailed research on why and
1169 how it occurs (http://curl.haxx.se/mail/lib-2009-04/0289.html). The
1170 conclusion was that if an error is detected and Curl_done() is called for
1171 the connection, ftp_done() could at times return another error code that
1172 then would take precedence and that new code confused existing logic that
1173 works for the first error code (CURLE_SEND_ERROR) only.
1175 - Gisle Vanem noticed that --libtool would produce bogus strings at times for
1176 OBJECTPOINT options. Now we've introduced a new function - my_setopt_str -
1177 within the app for setting plain string options to avoid the risk of this
1180 Daniel Stenberg (17 Apr 2009)
1181 - Pramod Sharma reported and tracked down a bug when doing FTP over a HTTP
1182 proxy. libcurl would then wrongly close the connection after each
1183 request. In his case it had the weird side-effect that it killed NTLM auth
1184 for the proxy causing an inifinite loop!
1186 I added test case 1098 to verify this fix. The test case does however not
1187 properly verify that the transfers are done persistently - as I couldn't
1188 think of a clever way to achieve it right now - but you need to read the
1189 stderr output after a test run to see that it truly did the right thing.
1191 Daniel Stenberg (13 Apr 2009)
1192 - bug report #2727981 (http://curl.haxx.se/bug/view.cgi?id=2727981) by Martin
1193 Storsjö pointed out how setting CURLOPT_NOBODY to 0 could be downright
1194 confusing as it set the method to either GET or HEAD. The example he showed
1197 curl_easy_setopt(curl, CURLOPT_PUT, 1);
1198 curl_easy_setopt(curl, CURLOPT_NOBODY, 0);
1200 The new way doesn't alter the method until the request is about to start. If
1201 CURLOPT_NOBODY is then 1 the HTTP request will be HEAD. If CURLOPT_NOBODY is
1202 0 and the request happens to have been set to HEAD, it will then instead be
1203 set to GET. I believe this will be less surprising to users, and hopefully
1204 not hit any existing users badly.
1206 - Toshio Kuratomi reported a memory leak problem with libcurl+NSS that turned
1207 out to be leaking cacerts. Kamil Dudka helped me complete the fix. The issue
1208 is found in Redhat's bug tracker:
1209 https://bugzilla.redhat.com/show_bug.cgi?id=453612
1211 There are still memory leaks present, but they seem to have other reasons.
1213 Daniel Fandrich (11 Apr 2009)
1214 - Added new libcurl source files to Symbian OS build files.
1215 - Improved Symbian support for SSL.
1217 Yang Tse (10 Apr 2009)
1218 - Daniel Johnson improved the MacOSX-Framework shell script to now perform all
1219 the steps required to build a Mac OS X four way fat ppc/i386/ppc64/x86_64
1220 libcurl.framework. Four way fat framework requires OS X 10.5 SDK or later.
1222 Yang Tse (8 Apr 2009)
1223 - Removed Sun compilers preprocessor block from curlbuild.h.dist, this also
1224 removes it from the curlbuild.h file originally distributed by the cURL
1225 project as this file is intended for systems not capable of running the
1226 configure script. For those who have been building curl out of the source
1227 code curl distribution tarball provided by curl.haxx.se the change implies
1228 nothing. Previous change in this area committed 2 Apr becomes irrelevant.
1230 Daniel Stenberg (6 Apr 2009)
1231 - I clarified in the docs that CURLOPT_SEEKFUNCTION should return 0 on success
1232 and 1 on fatal errors. Previously it only mentioned non-zero on fatal
1233 errors. This is a slight change in meaning, but it follows what we've done
1234 elsewhere before and it opens up for LOTS of more useful return codes
1235 whenever we can think of them...
1237 Yang Tse (2 Apr 2009)
1238 - Fix curl_off_t definition for builds done using Sun compilers and a
1239 non-configured libcurl. In this case curl_off_t data type was gated
1240 to the off_t data type which depends on the _FILE_OFFSET_BITS. This
1241 configuration is exactly the unwanted configuration for our curl_off_t
1242 data type which must not depend on such setting. This breaks ABI for
1243 libcurl libraries built with Sun compilers which were built without
1244 having run the configure script with _FILE_OFFSET_BITS different than
1245 64 and using the ILP32 data model.
1247 Daniel Stenberg (1 Apr 2009)
1248 - Andre Guibert de Bruet fixed a NULL pointer use in an infof() call if a
1249 strdup() call failed.
1251 Daniel Fandrich (31 Mar 2009)
1252 - Properly return an error code in curl_easy_recv (reported by Jim Freeman).
1254 Daniel Stenberg (18 Mar 2009)
1255 - Kamil Dudka brought a patch that enables 6 additional crypto algorithms when
1256 NSS is used. These ciphers were added in NSS 3.4 and require to be enabled
1259 Daniel Stenberg (13 Mar 2009)
1260 - Use libssh2_version() to present the libssh2 version in case the libssh2
1261 library is found to support it.
1263 Yang Tse (12 Mar 2009)
1264 - Added missing Curl_read() return code checking in TELNET transfers.
1266 - Pierre Brico found and fixed TELNET transfers not being aborted upon
1267 a write callback failure.
1269 Daniel Stenberg (11 Mar 2009)
1270 - Kamil Dudka made the curl tool properly call curl_global_init() before any
1271 other libcurl function.
1273 Yang Tse (11 Mar 2009)
1274 - Added missing TELNET timeout support for Windows builds. This issue was
1275 reported by Pierre Brico.
1277 Daniel Stenberg (9 Mar 2009)
1278 - Frank Hempel found out a bug and provided the fix:
1280 curl_easy_duphandle did not necessarily duplicate the CURLOPT_COOKIEFILE
1281 option. It only enabled the cookie engine in the destination handle if
1282 data->cookies is not NULL (where data is the source handle). In case of a
1283 newly initialized handle which just had the cookie support enabled by a
1284 curl_easy_setopt(handle, CURL_COOKIEFILE, "")-call, handle->cookies was
1285 still NULL because the setopt-call only appends the value to
1286 data->change.cookielist, hence duplicating this handle would not have the
1287 cookie engine switched on.
1289 We also concluded that the slist-functionality would be suitable for being
1290 put in its own module rather than simply hanging out in lib/sendf.c so I
1291 created lib/slist.[ch] for them.
1293 - Andreas Farber made the 'buildconf' script check for the presence of m4
1294 scripts to make it detect a bad checkout earlier. People with older
1295 checkouts who don't do cvs update with the -d option won't get the new dirs
1296 and then will get funny outputs that can be a bit hard to understand and
1299 Daniel Stenberg (8 Mar 2009)
1300 - Andre Guibert de Bruet found and fixed a code segment in ssluse.c where the
1301 allocation of the memory BIO was not being properly checked.
1303 - Andre Guibert de Bruet fixed the gnutls-using code: There are a few places
1304 in the gnutls code where we were checking for negative values for errors,
1305 when the man pages state that GNUTLS_E_SUCCESS is returned on success and
1306 other values indicate error conditions.
1308 - Bill Egert pointed out (http://curl.haxx.se/bug/view.cgi?id=2671602) that
1309 curl didn't use sprintf() in a way that is documented to work in POSIX but
1310 since we use our own printf() code (from libcurl) that shouldn't be a
1311 problem. Nonetheless I modified the code to not rely on such particular
1312 features and to not cause further raised eyebrowse with no good reason.
1314 Daniel Fandrich (5 Mar 2009)
1315 - Expanded the security section of the libcurl-tutorial man page to cover
1316 more issues for authors to consider when writing robust libcurl-using
1319 Yang Tse (5 Mar 2009)
1320 - Fixed NTLM authentication memory leak on SSPI enabled Windows builds. This
1321 issue was noticed by Chris Deidun.
1323 Daniel Fandrich (4 Mar 2009)
1324 - Fixed a problem with m4 quoting in the OpenSSL configure check reported
1327 Daniel Stenberg (3 Mar 2009)
1328 - David James brought a patch that make libcurl close (all) dead connections
1329 whenever you attempt to open a new connection.
1331 1. After cleaning up a dead connection, "continue" instead of
1332 returning FALSE. This ensures that we clean up all dead connections,
1333 rather than just cleaning up the first dead connection.
1334 2. Move up the cleanup for dead connections so that it occurs for
1335 all connections, rather than just the connections which have the same
1336 preferences as our current new connection.
1338 Version 7.19.4 (3 March 2009)
1340 Daniel Stenberg (3 Mar 2009)
1341 - David Kierznowski notified us about a security flaw
1342 (http://curl.haxx.se/docs/adv_20090303.html also known as CVE-2009-0037) in
1343 which previous libcurl versions (by design) can be tricked to access an
1344 arbitrary local/different file instead of a remote one when
1345 CURLOPT_FOLLOWLOCATION is enabled. This flaw is now fixed in this release
1346 together this the addition of two new setopt options for controlling this
1349 o CURLOPT_REDIR_PROTOCOLS controls what protocols libcurl is allowed to
1350 follow to when CURLOPT_FOLLOWLOCATION is enabled. By default, this option
1351 excludes the FILE and SCP protocols and thus you nee to explicitly allow
1352 them in your app if you really want that behavior.
1354 o CURLOPT_PROTOCOLS controls what protocol(s) libcurl is allowed to fetch
1355 using the primary URL option. This is useful if you want to allow a user or
1356 other outsiders control what URL to pass to libcurl and yet not allow all
1357 protocols libcurl may have been built to support.
1359 Daniel Stenberg (27 Feb 2009)
1360 - Senthil Raja Velu reported a problem when CURLOPT_INTERFACE and
1361 CURLOPT_LOCALPORT were used together (the local port bind failed), and
1362 Markus Koetter provided the fix!
1364 Daniel Stenberg (25 Feb 2009)
1365 - As Daniel Fandrich figured out, we must do the GnuTLS initing in the
1366 curl_global_init() function to properly maintain the performing functions
1367 thread-safe. We've previously (28 April 2007) moved the init to a later time
1368 just to avoid it to fail very early when libgcrypt dislikes the situation,
1369 but that move was bad and the fix should rather be in libgcrypt or
1372 Daniel Stenberg (24 Feb 2009)
1373 - Brian J. Murrell found out that Negotiate proxy authentication didn't work.
1374 It happened because the code used the struct for server-based auth all the
1375 time for both proxy and server auth which of course was wrong.
1377 Daniel Stenberg (23 Feb 2009)
1378 - After a bug reported by James Cheng I've made curl_easy_getinfo() for
1379 CURLINFO_CONTENT_LENGTH_DOWNLOAD and CURLINFO_CONTENT_LENGTH_UPLOAD return
1380 -1 if the sizes aren't know. Previously these returned 0, make it impossible
1381 to detect the difference between actually zero and unknown.
1383 Yang Tse (23 Feb 2009)
1384 - Daniel Johnson provided a shell script that will perform all the steps needed
1385 to build a Mac OS X fat ppc/i386 or ppc64/x86_64 libcurl.framework
1387 Daniel Stenberg (23 Feb 2009)
1388 - I renamed everything in the windows builds files that used the name 'curllib'
1389 to the proper 'libcurl' as clearly this caused confusion.
1391 Yang Tse (20 Feb 2009)
1392 - Do not halt compilation when using VS2008 to build a Windows 2000 target.
1394 Daniel Stenberg (20 Feb 2009)
1395 - Linus Nielsen Feltzing reported and helped me repeat and fix a problem with
1396 FTP with the multi interface: when a transfer fails, like when aborted by a
1397 write callback, the control connection was wrongly closed and thus not
1400 This change is also an attempt to cleanup the code somewhat in this area, as
1401 now the FTP code attempts to keep (better) track on pending responses
1402 necessary to get read in ftp_done().
1404 Daniel Stenberg (19 Feb 2009)
1405 - Patrik Thunstrom reported a problem and helped me repeat it. It turned out
1406 libcurl did a superfluous 1000ms wait when doing SFTP downloads!
1408 We read data with libssh2 while doing the "DO" operation for SFTP and then
1409 when we were about to start getting data for the actual file part, the
1410 "TRANSFER" part, we waited for socket action (in 1000ms) before doing a
1411 libssh2-read. But in this case libssh2 had already read and buffered the
1412 data so we ended up always just waiting 1000ms before we get working on the
1415 Patrick Monnerat (18 Feb 2009)
1416 - FTP downloads (i.e.: RETR) ending with code 550 now return error
1417 CURLE_REMOTE_FILE_NOT_FOUND instead of CURLE_FTP_COULDNT_RETR_FILE.
1419 Daniel Stenberg (17 Feb 2009)
1420 - Kamil Dudka made NSS-powered builds compile and run again!
1422 - A second follow-up change by Andre Guibert de Bruet to fix a related memory
1423 leak like that fixed on the 14th. When zlib returns failure, we need to
1424 cleanup properly before returning error.
1426 - CURLOPT_FTP_CREATE_MISSING_DIRS can now be set to 2 in addition to 1 for
1427 plain FTP connections, and it will then allow MKD to fail once and retry the
1428 CWD afterwards. This is especially useful if you're doing many simultanoes
1429 connections against the same server and they all have this option enabled,
1430 as then CWD may first fail but then another connection does MKD before this
1431 connection and thus MKD fails but trying CWD works! The numbers can
1432 (should?) now be set with the convenience enums now called
1433 CURLFTP_CREATE_DIR and CURLFTP_CREATE_DIR_RETRY.
1435 Tests has proven that if you're making an application that uploads a set of
1436 files to an ftp server, you will get a noticable gain in speed if you're
1437 using multiple connections and this option will be then be very useful.
1439 Daniel Stenberg (14 Feb 2009)
1440 - Andre Guibert de Bruet found and fixed a memory leak in the content encoding
1441 code, which could happen on libz errors.
1443 Daniel Fandrich (12 Feb 2009)
1444 - Added support for Digest and NTLM authentication using GnuTLS.
1446 Daniel Stenberg (11 Feb 2009)
1447 - CURLINFO_CONDITION_UNMET was added to allow an application to get to know if
1448 the condition in the previous request was unmet. This is typically a time
1449 condition set with CURLOPT_TIMECONDITION and was previously not possible to
1450 reliably figure out. From bug report #2565128
1451 (http://curl.haxx.se/bug/view.cgi?id=2565128) filed by Jocelyn Jaubert.
1453 Daniel Fandrich (4 Feb 2009)
1454 - Don't add the standard /usr/lib or /usr/include paths to LDFLAGS and CPPFLAGS
1455 (respectively) when --with-ssl=/usr is used (patch based on FreeBSD).
1457 - Added an explicit buffer limit check in msdosify() (patch based on FreeBSD).
1458 This couldn't ever overflow in curl, but might if the code were used
1459 elsewhere or under different conditions.
1461 Daniel Stenberg (3 Feb 2009)
1462 - Hidemoto Nakada provided a small fix that makes it possible to get the
1463 CURLINFO_CONTENT_LENGTH_DOWNLOAD size from file:// "transfers" with
1464 CURLOPT_NOBODY set true.
1466 Daniel Stenberg (2 Feb 2009)
1467 - Patrick Scott found a rather large memory leak when using the multi
1468 interface and setting CURLMOPT_MAXCONNECTS to something less than the number
1469 of handles you add to the multi handle. All the connections that didn't fit
1470 in the cache would not be properly disconnected nor freed!
1472 - Craig A West brought us: libcurl now defaults to do CONNECT with HTTP
1473 version 1.1 instead of 1.0 like before. This change also introduces the new
1474 proxy type for libcurl called 'CURLPROXY_HTTP_1_0' that then allows apps to
1475 switch (back) to CONNECT 1.0 requests. The curl tool also got a --proxy1.0
1476 option that works exactly like --proxy but sets CURLPROXY_HTTP_1_0.
1478 I updated all test cases cases that use CONNECT and I tried to do some using
1479 --proxy1.0 and some updated to do CONNECT 1.1 to get both versions run.
1481 Daniel Stenberg (31 Jan 2009)
1482 - When building with c-ares 1.6.1 (not yet released) or later and IPv6 support
1483 enabled, we can now take advantage of its brand new AF_UNSPEC support in
1484 ares_gethostbyname(). This makes test case 241 finally run fine for me with
1485 this setup since it now parses the "::1 ip6-localhost" line fine in my
1488 Daniel Stenberg (30 Jan 2009)
1489 - Scott Cantor filed bug report #2550061
1490 (http://curl.haxx.se/bug/view.cgi?id=2550061) mentioning that I failed to
1491 properly make sure that the VC9 makefiles got included in the latest
1492 release. I've now fixed the release script and verified it so next release
1493 will hopefully include them properly!
1495 Daniel Fandrich (30 Jan 2009)
1496 - Fixed --disable-proxy for FTP and SOCKS. Thanks to Daniel Egger for
1499 Yang Tse (29 Jan 2009)
1500 - Introduced curl_sspi.c and curl_sspi.h for the implementation of functions
1501 Curl_sspi_global_init() and Curl_sspi_global_cleanup() which previously were
1502 named Curl_ntlm_global_init() and Curl_ntlm_global_cleanup() in http_ntlm.c
1503 Also adjusted socks_sspi.c to remove the link-time dependency on the Windows
1504 SSPI library using it now in the same way as it was done in http_ntlm.c.
1506 Daniel Stenberg (28 Jan 2009)
1507 - Markus Moeller introduced two new options to libcurl:
1508 CURLOPT_SOCKS5_GSSAPI_SERVICE and CURLOPT_SOCKS5_GSSAPI_NEC to allow libcurl
1509 to do GSS-style authentication with SOCKS5 proxies. The curl tool got the
1510 options called --socks5-gssapi-service and --socks5-gssapi-nec to enable
1513 Daniel Stenberg (26 Jan 2009)
1514 - Chad Monroe provided the new CURLOPT_TFTP_BLKSIZE option that allows an app
1515 to set desired block size to use for TFTP transfers instead of the default
1518 - The "-no_ticket" option was introduced in Openssl0.9.8j. It's a flag to
1519 disable "rfc4507bis session ticket support". rfc4507bis was later turned
1520 into the proper RFC5077 it seems: http://tools.ietf.org/html/rfc5077
1522 The enabled extension concerns the session management. I wonder how often
1523 libcurl stops a connection and then resumes a TLS session. also, sending the
1524 session data is some overhead. .I suggest that you just use your proposed
1525 patch (which explicitly disables TICKET).
1527 If someone writes an application with libcurl and openssl who wants to
1528 enable the feature, one can do this in the SSL callback.
1530 Sharad Gupta brought this to my attention. Peter Sylvester helped me decide
1531 on the proper action.
1533 - Alexey Borzov filed bug report #2535504
1534 (http://curl.haxx.se/bug/view.cgi?id=2535504) pointing out that realms with
1535 quoted quotation marks in HTTP Digest headers didn't work. I've now added
1536 test case 1095 that verifies my fix.
1538 - Craig A West brought CURLOPT_NOPROXY and the corresponding --noproxy option.
1539 They basically offer the same thing the NO_PROXY environment variable only
1540 offered previously: list a set of host names that shall not use the proxy
1541 even if one is specified.
1543 Daniel Fandrich (20 Jan 2009)
1544 - Call setlocale() for libtest tests to test the effects of locale-induced
1545 libc changes on libcurl.
1547 - Fixed a couple more locale-dependent toupper conversions, mainly for
1548 clarity. This does fix one problem that causes ;type=i FTP URLs
1549 to fail in the Turkish locale when CURLOPT_PROXY_TRANSFER_MODE is
1550 used (test case 561)
1552 - Added tests 561 and 1091 through 1094 to test various combinations
1553 of ;type= and ;mode= URLs that could potentially fail in the Turkish
1556 Daniel Stenberg (20 Jan 2009)
1557 - Lisa Xu pointed out that the ssh.obj file was missing from the
1558 lib/Makefile.vc6 file (and thus from the vc8 and vc9 ones too).
1560 Version 7.19.3 (19 January 2009)
1562 Daniel Stenberg (16 Jan 2009)
1563 - Andrew de los Reyes fixed curlbuild.h for "generic" gcc builds on PPC, both
1566 Daniel Stenberg (15 Jan 2009)
1567 - Tim Ansell fixed a compiler warning in lib/cookie.c
1569 Daniel Stenberg (14 Jan 2009)
1570 - Grant Erickson fixed timeouts for TFTP such that specifying a
1571 connect-timeout, a max-time or both options work correctly and as expected
1572 by passing the correct boolean value to Curl_timeleft via the
1573 'duringconnect' parameter.
1575 With this small change, curl TFTP now behaves as expected (and likely as
1576 originally-designed):
1578 1) For non-existent or unreachable dotted IP addresses:
1580 a) With no options, follows the default curl 300s timeout...
1581 b) With --connect-timeout only, follows that value...
1582 c) With --max-time only, follows that value...
1583 d) With both --connect-timeout and --max-time, follows the smaller value...
1585 and times out with a "curl: (7) Couldn't connect to server" error.
1587 2) For transfers to/from a valid host:
1589 a) With no options, follows default curl 300s timeout for the
1590 first XRQ/DATA/ACK transaction and the default TFTP 3600s
1591 timeout for the remainder of the transfer...
1593 b) With --connect-time only, follows that value for the
1594 first XRQ/DATA/ACK transaction and the default TFTP 3600s
1595 timeout for the remainder of the transfer...
1597 c) With --max-time only, follows that value for the first
1598 XRQ/DATA/ACK transaction and for the remainder of the
1601 d) With both --connect-timeout and --max-time, follows the former
1602 for the first XRQ/DATA/ACK transaction and the latter for the
1603 remainder of the transfer...
1605 and times out with a "curl: (28) Timeout was reached" error as
1608 Daniel Stenberg (13 Jan 2009)
1609 - Michael Wallner fixed a NULL pointer deref when calling
1610 curl_easy_setup(curl, CURLOPT_COOKIELIST, "SESS") on a CURL handle with no
1613 - Stefan Teleman brought a patch to fix the default curlbuild.h file for the
1616 Daniel Stenberg (12 Jan 2009)
1617 - Based on bug report #2498665 (http://curl.haxx.se/bug/view.cgi?id=2498665)
1618 by Daniel Black, I've now added magic to the configure script that makes it
1619 use pkg-config to detect gnutls details as well if the existing method
1620 (using libgnutls-config) fails. While doing this, I cleaned up and unified
1621 the pkg-config usage when detecting openssl and nss as well.
1623 Daniel Stenberg (11 Jan 2009)
1624 - Karl Moerder brought the patch that creates vc9 Makefiles, and I made
1625 'maketgz' now use the actual makefile targets to do the VC8 and VC9
1628 Daniel Stenberg (10 Jan 2009)
1629 - Emil Romanus fixed:
1631 When using the multi interface over HTTP and the server returns a Location
1632 header, the running easy handle will get stuck in the CURLM_STATE_PERFORM
1633 state, leaving the external event loop stuck waiting for data from the
1634 ingoing socket (when using the curl_multi_socket_action stuff). While this
1635 bug was pretty hard to find, it seems to require only a one-line fix. The
1636 break statement on line 1374 in multi.c caused the function to skip the call
1639 How to reproduce this bug? Well, that's another question. evhiperfifo.c in
1640 the examples directory chokes on this bug only _sometimes_, probably
1641 depending on how fast the URLs are added. One way of testing the bug out is
1642 writing to hiper.fifo from more than one source at the same time.
1644 Daniel Fandrich (7 Jan 2009)
1645 - Unified much of the SessionHandle initialization done in Curl_open() and
1646 curl_easy_reset() by creating Curl_init_userdefined(). This had the side
1647 effect of fixing curl_easy_reset() so it now also resets
1648 CURLOPT_FTP_FILEMETHOD and CURLOPT_SSL_SESSIONID_CACHE
1650 Daniel Stenberg (7 Jan 2009)
1651 - Rob Crittenden did once again provide an NSS update:
1653 I have to jump through a few hoops now with the NSS library initialization
1654 since another part of an application may have already initialized NSS by the
1655 time Curl gets invoked. This patch is more careful to only shutdown the NSS
1656 library if Curl did the initialization.
1658 It also adds in a bit of code to set the default ciphers if the app that
1659 call NSS_Init* did not call NSS_SetDomesticPolicy() or set specific
1660 ciphers. One might argue that this lets other application developers get
1661 lazy and/or they aren't using the NSS API correctly, and you'd be right.
1662 But still, this will avoid terribly difficult-to-trace crashes and is
1665 Daniel Stenberg (1 Jan 2009)
1666 - 'reconf' is removed since we rather have users use 'buildconf'
1668 Daniel Stenberg (31 Dec 2008)
1669 - Bas Mevissen reported http://curl.haxx.se/bug/view.cgi?id=2479030 pointing
1670 out that 'reconf' didn't properly point out the m4 subdirectory when running
1673 Daniel Stenberg (29 Dec 2008)
1674 - Phil Lisiecki filed bug report #2413067
1675 (http://curl.haxx.se/bug/view.cgi?id=2413067) that identified a problem that
1676 would cause libcurl to mark a DNS cache entry "in use" eternally if the
1677 subsequence TCP connect failed. It would thus never get pruned and refreshed
1678 as it should've been.
1680 Phil provided his own patch to this problem that while it seemed to work
1681 wasn't complete and thus I wrote my own fix to the problem.
1683 Daniel Stenberg (28 Dec 2008)
1684 - Peter Korsgaard fixed building libcurl with "configure --with-ssl
1687 - Anthony Bryan fixed more language and spelling flaws in man pages.
1689 Daniel Stenberg (22 Dec 2008)
1690 - Given a recent enough libssh2, libcurl can now seek/resume with SFTP even
1691 on file indexes beyond 2 or 4GB.
1693 - Anthony Bryan provided a set of patches that cleaned up manual language,
1694 corrected spellings and more.
1696 Daniel Stenberg (20 Dec 2008)
1697 - Igor Novoseltsev fixed a bad situation for the multi_socket() API when doing
1698 pipelining, as libcurl could then easily get confused and A) work on the
1699 handle that was not "first in queue" on a pipeline, or even B) tell the app
1700 to REMOVE a socket while it was in use by a second handle in a pipeline. Both
1701 errors caused hanging or stalling applications.
1703 Daniel Stenberg (19 Dec 2008)
1704 - curl_multi_timeout() could return a timeout value of 0 even though nothing
1705 was actually ready to get done, as the internal time resolution is higher
1706 than the returned millisecond timer. Therefore it could cause applications
1707 running on fast processors to do short bursts of busy-loops.
1708 curl_multi_timeout() will now only return 0 if the timeout is actually
1711 - Using the libssh2 0.19 function libssh2_session_block_directions(), libcurl
1712 now has an improved ability to do right when the multi interface (both
1713 "regular" and multi_socket) is used for SCP and SFTP transfers. This should
1714 result in (much) less busy-loop situations and thus less CPU usage with no
1717 Daniel Stenberg (17 Dec 2008)
1718 - SCP and SFTP with the multi interface had the same flaw: the 'DONE'
1719 operation didn't complete properly if the EAGAIN equivalent was returned but
1720 libcurl would simply continue with a half-completed close operation
1721 performed. This ruined persistent connection re-use and cause some
1722 SSH-protocol errors in general. The correction is unfortunately adding a
1723 blocking function - doing it entirely non-blocking should be considered for
1726 Gisle Vanem (16 Dec 2008)
1727 - Added the possibility to use the Watt-32 tcp/ip stack under Windows.
1728 The change simply involved adding a USE_WATT32 section in the
1729 config-win32.h files (under ./lib and ./src). This section disables
1730 the use of any Winsock headers.
1732 Daniel Stenberg (16 Dec 2008)
1733 - libssh2_sftp_last_error() was wrongly used at some places in libcurl which
1734 made libcurl sometimes not properly abort problematic SFTP transfers.
1736 Daniel Stenberg (12 Dec 2008)
1737 - More work with Igor Novoseltsev to first fix the remaining stuff for
1738 removing easy handles from multi handles when the easy handle is/was within
1739 a HTTP pipeline. His bug report #2351653
1740 (http://curl.haxx.se/bug/view.cgi?id=2351653) was also related and was
1741 eventually fixed by a patch by Igor himself.
1743 Yang Tse (12 Dec 2008)
1744 - Patrick Monnerat fixed a build regression, introduced in 7.19.2, affecting
1745 OS/400 compilations with IPv6 enabled.
1747 Daniel Stenberg (12 Dec 2008)
1748 - Mark Karpeles filed bug report #2416182 titled "crash in ConnectionExists
1749 when using duphandle+curl_mutli"
1750 (http://curl.haxx.se/bug/view.cgi?id=2416182) which showed that
1751 curl_easy_duphandle() wrongly also copied the pointer to the connection
1752 cache, which was plain wrong and caused a segfault if the handle would be
1753 used in a different multi handle than the handle it was duplicated from.
1755 Daniel Stenberg (11 Dec 2008)
1756 - Keshav Krity found out that libcurl failed to deal with dotted IPv6
1757 addresses if they were very long (>39 letters) due to a too strict address
1758 validity parser. It now accepts addresses up to 45 bytes long.
1760 Daniel Stenberg (11 Dec 2008)
1761 - Internet Explorer had a broken HTTP digest authentication before v7 and
1762 there are servers "out there" that relies on the client doing this broken
1763 Digest authentication. Apache even comes with an option to work with such
1766 The difference is only for URLs that contain a query-part (a '?'-letter and
1767 text to the right of it).
1769 libcurl now supports this quirk, and you enable it by setting the
1770 CURLAUTH_DIGEST_IE bit in the bitmask you pass to the CURLOPT_HTTPAUTH or
1771 CURLOPT_PROXYAUTH options. They are thus individually controlled to server
1774 (note that there's no way to activate this with the curl tool yet)
1776 Daniel Fandrich (9 Dec 2008)
1777 - Added test cases 1089 and 1090 to test --write-out after a redirect to
1778 test a report that the size didn't work, but these test cases pass.
1780 - Documented CURLOPT_CONNECT_ONLY as being useful only on HTTP URLs.
1782 Daniel Stenberg (9 Dec 2008)
1783 - Ken Hirsch simplified how libcurl does FTPS: now it doesn't assume any
1784 particular state for the control connection like it did before for implicit
1785 FTPS (libcurl assumed such control connections to be encrypted while some
1786 FTPS servers such as FileZilla assumes such connections to be clear
1787 mode). Use the CURLOPT_USE_SSL option to set your desired level.
1789 Daniel Stenberg (8 Dec 2008)
1790 - Fred Machado posted about a weird FTP problem on the curl-users list and when
1791 researching it, it turned out he got a 550 response back from a SIZE command
1792 and then I fell over the text in RFC3659 that says:
1794 The presence of the 550 error response to a SIZE command MUST NOT be taken
1795 by the client as an indication that the file cannot be transferred in the
1796 current MODE and TYPE.
1798 In other words: the change I did on September 30th 2008 and that has been
1799 included in the last two releases were a regression and a bad idea. We MUST
1800 NOT take a 550 response from SIZE as a hint that the file doesn't exist.
1802 - Christian Krause filed bug #2221237
1803 (http://curl.haxx.se/bug/view.cgi?id=2221237) that identified an infinite
1804 loop during GSS authentication given some specific conditions. With his
1805 patience and great feedback I managed to narrow down the problem and
1806 eventually fix it although I can't test any of this myself!
1808 Daniel Fandrich (3 Dec 2008)
1809 - Fixed the getifaddrs version of Curl_if2ip to work on systems without IPv6
1810 support (e.g. Minix)
1812 Daniel Stenberg (3 Dec 2008)
1813 - Igor Novoseltsev filed bug #2351645
1814 (http://curl.haxx.se/bug/view.cgi?id=2351645) that identified a problem with
1815 the multi interface that occured if you removed an easy handle while in
1816 progress and the handle was used in a HTTP pipeline.
1818 - Pawel Kierski pointed out a mistake in the cookie code that could lead to a
1819 bad fclose() after a fatal error had occured.
1820 (http://curl.haxx.se/bug/view.cgi?id=2382219)
1822 Daniel Fandrich (25 Nov 2008)
1823 - If a HTTP request is Basic and num is already >=1000, the HTTP test
1824 server adds 1 to num to get the data section to return. This allows
1825 testing authentication negotiations using the Basic authentication
1828 - Added tests 1087 and 1088 to test Basic authentication on a redirect
1829 with and without --location-trusted
1831 Daniel Stenberg (24 Nov 2008)
1832 - Based on a patch by Vlad Grachov, libcurl now uses a new libssh2 0.19
1833 function when built to support SCP and SFTP that helps the library to know
1834 in which direction a particular libssh2 operation would return EAGAIN so
1835 that libcurl knows what socket conditions to wait for before trying the
1836 function call again. Previously (and still when using libssh2 0.18 or
1837 earlier), libcurl will busy-loop in this situation when the easy interface
1840 Daniel Fandrich (20 Nov 2008)
1841 - Automatically detect OpenBSD's CA cert bundle.
1843 Daniel Stenberg (19 Nov 2008)
1844 - I removed the default use of "Pragma: no-cache" from libcurl when a proxy is
1845 used. It has been used since forever but it was never a good idea to use
1846 unless explicitly asked for.
1848 - Josef Wolf's extension that allows a $TESTDIR/gdbinit$testnum file that when
1849 you use runtests.pl -g, will be sourced by gdb to allow additional fancy or
1850 whatever you see fit
1852 - Christian Krause reported and fixed a memory leak that would occur with HTTP
1853 GSS/kerberos authentication (http://curl.haxx.se/bug/view.cgi?id=2284386)
1855 - Andreas Wurf and Markus Koetter helped me analyze a problem that Andreas got
1856 when uploading files to a single FTP server using multiple easy handle
1857 handles with the multi interface. Occasionally a handle would stall in
1860 The problem turned out to be a side-effect of the ConnectionExists()
1861 function's eagerness to re-use a handle for HTTP pipelining so it would
1862 select it even if already being in use, due to an inadequate check for its
1863 chances of being used for pipelnining.
1865 Daniel Fandrich (17 Nov 2008)
1866 - Added more compiler warning options for gcc 4.3
1868 Yang Tse (17 Nov 2008)
1869 - Fix a remaining problem in the inet_pton() runtime configure check. And
1870 fix internal Curl_inet_pton() failures to reject certain malformed literals.
1872 - Make configure script check if ioctl with the SIOCGIFADDR command can be
1873 used, and define HAVE_IOCTL_SIOCGIFADDR if appropriate.
1875 Daniel Stenberg (16 Nov 2008)
1876 - Christian Krause fixed a build failure when building with gss support
1877 enabled and FTP disabled.
1879 - Added check for NULL returns from strdup() in src/main.c and lib/formdata.c
1880 - reported by Jim Meyering also prevent buffer overflow on MSDOS when you do
1881 for example -O on a url with a file name part longer than PATH_MAX letters
1883 - lib/nss.c fixes based on the report by Jim Meyering: I went over and added
1884 checks for return codes for all calls to malloc and strdup that were
1885 missing. I also changed a few malloc(13) to use arrays on the stack and a
1886 few malloc(PATH_MAX) to instead use aprintf() to lower memory use.
1888 - I fixed a memory leak in Curl_nss_connect() when CURLOPT_ISSUERCERT is
1891 Daniel Fandrich (14 Nov 2008)
1892 - Added .xml as one of the few common file extensions known by the multipart
1895 - Added some #ifdefs around header files and change the EAGAIN test to
1896 fix compilation on Cell (reported by Jeff Curley).
1898 Yang Tse (14 Nov 2008)
1899 - Fixed several configure script issues affecting checks for inet_ntoa_r(),
1900 inet_ntop(), inet_pton(), getifaddrs(), fcntl() and getaddrinfo().
1902 Yang Tse (13 Nov 2008)
1903 - Refactored configure script detection of functions used to set sockets into
1904 non-blocking mode, and decouple function detection from function capability.