5 \___|\___/|_| \_\_____|
9 Daniel Stenberg (14 May 2010)
10 - John-Mark Bell filed bug #3000052 that identified a problem (with an
11 associated patch) with the OpenSSL handshake state machine when the multi
14 Performing an https request using a curl multi handle and using select or
15 epoll to wait for events results in a hang. It appears that the cause is the
16 fix for bug #2958179, which makes ossl_connect_common unconditionally return
17 from the step 2 loop when fetching from a multi handle.
19 When ossl_connect_step2 has completed, it updates connssl->connecting_state
20 to ssl_connect_3. ossl_connect_common will then return to the caller, as a
21 multi handle is in use. Eventually, the client code will call
22 curl_multi_fdset to obtain an updated fdset to select or epoll on. For https
23 requests, curl_multi_fdset will cause https_getsock to be called.
24 https_getsock will only return a socket handle if the connecting_state is
25 ssl_connect_2_reading or ssl_connect_2_writing. Therefore, the client will
26 never obtain a valid fdset, and thus not drive the multi handle, resulting
29 (http://curl.haxx.se/bug/view.cgi?id=3000052)
31 - Sebastian V reported bug #3000056 identifying a problem with redirect
32 following. It showed that when curl followed redirects it didn't properly
33 ignore the response body of the 30X response if that response was using
34 compressed Content-Encoding!
36 (http://curl.haxx.se/bug/view.cgi?id=3000056)
38 Daniel Stenberg (12 May 2010)
39 - Howard Chu brought support for RTMP. This is powered by the underlying
40 librtmp library. It supports a range of variations and "sub-protocols"
41 within the RTMP family.
43 - Pavel Raiskup brought support for FTP directory wildcard matching to allow
44 selective downloading. To provide that, a set of new options were added:
47 CURLOPT_CHUNK_BGN_FUNCTION
48 CURLOPT_CHUNK_END_FUNCTION
50 CURLOPT_FNMATCH_FUNCTION
52 There were also a set of new tests added (574 - 577) to verify this.
54 Kamil Dudka (11 May 2010)
55 - CRL support in libcurl-NSS has been completely broken. Now it works. Original
56 bug report: https://bugzilla.redhat.com/581926
58 Daniel Stenberg (7 May 2010)
59 - Dirk Manske reported a regression. When connecting with the multi interface,
60 there were situations where libcurl wouldn't store connect time correctly as
61 it used to (and is documented to) do.
63 Using his fine sample program we could repeat it, and I wrote up test case
64 573 using that code. The problem does not easily show itself using the local
67 The fix, also as suggested by Dirk, is a bit on the ugly side as it adds yet
68 another call to Curl_verboseconnect() and setting the TIMER_CONNECT time.
69 That situation is subject for some closer inspection in the future.
71 - Howard Chu split the I/O handling functions into private handlers.
73 Howard Chu brought the bulk work of this patch that properly moves out the
74 sending and recving of data to the parts of the code that are properly
75 responsible for the various ways of doing so.
77 Daniel Stenberg assisted with polishing a few bits and fixed some minor
78 flaws in the original patch.
80 Another upside of this patch is that we now abuse CURLcodes less with the
81 "magic" -1 return codes and instead use CURLE_AGAIN more consistently.
83 Daniel Stenberg (5 May 2010)
84 - Hoi-Ho Chan introduced support for using the PolarSSL library. You control
85 this with the new configure option --with-polarssl.
87 Daniel Stenberg (29 Apr 2010)
88 - Ben Greear made telnet a lot better/easier to use by an application:
90 The main change is to allow input from user-specified methods, when they are
91 specified with CURLOPT_READFUNCTION. All calls to fflush(stdout) in
92 telnet.c were removed, which makes using 'curl telnet://foo.com' painful
93 since prompts and other data are not always returned to the user promptly.
94 Use 'curl --no-buffer telnet://foo.com' instead. In general, the user
95 should have their CURLOPT_WRITEFUNCTION do a fflush for interactive use.
97 Also fix assumption that reading from stdin never returns < 0.
98 Old code could crash in that case.
100 Call progress functions in telnet main loop.
102 Daniel Stenberg (26 Apr 2010)
103 - Make use of the libssh2_init/exit functions that libssh2 added in version
104 1.2.5. Using them will improve how libcurl works in threaded situations when
105 SCP and SFTP are transfered.
107 Daniel Stenberg (25 Apr 2010)
108 - Based on work by Kamil Dudka, I've introduced the new configure option
109 --enable-threaded-resolver. When used, the configure script will check for
110 pthreads and if around, it will build libcurl to use pthreads to do name
111 resolving in a threaded manner. Note that this is just a fix to offer an
112 option that can enable the code that already included. The threader resolver
113 code was mostly added on Jan 26 2010.
115 Daniel Stenberg (24 Apr 2010)
116 - Alex Bligh introduced the --proto and -proto-redir options that limit what
117 protocols curl accepts for the requests and when following redirects.
119 Kamil Dudka (24 Apr 2010)
120 - Fixed test536 in order to not fail with threaded DNS resolver and tweaked
121 comments in certain examples using curl_multi_fdset().
123 - Fixed SSL handshake timeout underflow in libcurl-NSS, which caused test405
124 to hang on a slow machine.
126 Daniel Stenberg (21 Apr 2010)
127 - The -O option caused curl to crash on windows and DOS due to the tool
128 writing out of boundary memory.
130 Yang Tse (20 Apr 2010)
131 - Ruslan Gazizov detected that MSVC makefiles were using wsock32.lib instead
132 of ws2_32.lib, this generated linking issues on MSVC IPv6 enabled builds
133 that were done using those makefiles.
135 Daniel Stenberg (19 Apr 2010)
136 - -J/--remote-header-name didn't strip trailing carriage returns or linefeeds
137 properly, so they could be used in the file name.
139 Daniel Stenberg (16 Apr 2010)
140 - Jerome Vouillon made the GnuTLS SSL handshake phase non-blocking.
142 - The recent overhaul of the SSL recv function made the GnuTLS specific code
143 treat a zero returned from gnutls_record_recv() as an error, and this caused
144 our HTTPS test cases to fail. We leave it to upper layer code to detect if
145 an EOF is a problem or not.
147 - I reverted the resolver fix from yesterday and instead removed all uses of
148 AI_CANONNAME all over libcurl and made the only user of that info (krb5.c)
149 use the host name from the URL instead. No reverse resolving is a good
152 - Paul Howarth made configure properly detect GSS "on ancient Linux distros"
153 by editing in which order we use headers to detect GSS.
155 Daniel Stenberg (15 Apr 2010)
156 - Rainer Canavan filed bug report #2987196 that identified libcurl doing
157 unnecesary reverse name lookups in many cases when built to use IPv4 and
158 getaddrinfo(). The logic for ipv6 is now used for ipv4 too.
160 (http://curl.haxx.se/bug/view.cgi?id=2963679)
162 Version 7.20.1 (14 April 2010)
164 Daniel Stenberg (9 Apr 2010)
165 - Prefixing the FTP quote commands with an asterisk really only worked for the
166 postquote actions. This is now fixed and test case 227 has been extended to
169 Kamil Dudka (4 Apr 2010)
170 - Eliminated a race condition in Curl_resolv_timeout().
172 - Refactorized interface of Curl_ssl_recv()/Curl_ssl_send().
174 - libcurl-NSS now provides more accurate messages and error codes in case of
175 client certificate problem. Either during connection, or transfer phase.
177 Daniel Stenberg (1 Apr 2010)
178 - Matt Wixson found and fixed a bug in the SCP/SFTP area where the code
179 treated a 0 return code from libssh2 to be the same as EAGAIN while in
180 reality it isn't. The problem caused a hang in SFTP transfers from a
183 Daniel Stenberg (28 Mar 2010)
184 - Ben Greear: If you pass a URL to pop3 that does not contain a message ID as
185 part of the URL, it would previously ask for 'INBOX' which just causes the
186 pop3 server to return an error.
188 Now libcurl treats en empty message ID as a request for LIST (list of pop3
189 message IDs). User's code could then parse this and download individual
192 Daniel Stenberg (27 Mar 2010)
193 - Ben Greear brought a patch that from now on allows all protocols to specify
194 name and user within the URL, in the same manner HTTP and FTP have been
195 allowed to in the past - although far from all of the libcurl supported
196 protocls actually have that feature in their URL definition spec.
198 Daniel Stenberg (26 Mar 2010)
199 - Ben Greear brought code that makes the rate limiting code for the easy
200 interface a bit smoother as it introduces sub-second sleeps during it and it
201 also takes the buffer sizes into account.
203 Daniel Stenberg (24 Mar 2010)
204 - Bob Richmond: There's an annoying situation where libcurl will read new HTTP
205 response data from a socket, then check if it's a timeout if one is set. If
206 the last packet received constitutes the end of the response body, libcurl
207 still treats it as a timeout condition and reports a message like:
209 "Operation timed out after 3000 milliseconds with 876 out of 876 bytes
212 It should only a timeout if the timer lapsed and we DIDN'T receive the end
213 of the response body yet.
215 - Christopher Conroy fixed a problem with RTSP and GET_PARAMETER reported
216 to us by Massimo Callegari. There's a new test case 572 that verifies this
219 - The 'ares' subtree has been removed from the source repository. It was
220 always a separate project that sort of piggybacked on the curl project since
221 the dawn of times and now the time has come for it to go stand on its own
222 legs and continue living its own life. All details on c-ares and its new
223 source code repository is found at http://c-ares.haxx.se/
225 Daniel Stenberg (23 Mar 2010)
226 - Kenny To filed the bug report #2963679 with patch to fix a problem he
227 experienced with doing multi interface HTTP POST over a proxy using
228 PROXYTUNNEL. He found a case where it would connect fine but bits.tcpconnect
229 was not set correct so libcurl didn't work properly.
231 (http://curl.haxx.se/bug/view.cgi?id=2963679)
233 - Akos Pasztory filed debian bug report #572276
234 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=572276 mentioning a problem
235 with a resource that returns chunked-encoded _and_ with a Content-Length
236 and libcurl failed to properly ignore the latter information.
238 - Hauke Duden provided an example program that made the multi interface crash.
239 His example simply used the multi interface and did first one FTP transfer
240 and after completion it used a second easy handle and did another FTP
241 transfer on the same FTP server.
243 This triggered a bug in the "delayed easy handle kill" system that curl
244 uses: when an FTP connection is left alive it must keep an easy handle
245 around internally - only for the purpose of having an easy handle when it
246 later disconnects it. The code assumed that when the easy handle was removed
247 and an internal reference was made, that version could be killed later on
248 when a new easy handle came using the same connection. This was wrong as
249 Hauke's example showed that the removed handle wasn't killed for real until
250 later. This caused a double close attempt => segfault.
252 Daniel Stenberg (22 Mar 2010)
253 - Thomas Lopatic fixed the alarm()-based DNS timeout:
255 Looking at the code of Curl_resolv_timeout() in hostip.c, I think that in
256 case of a timeout, the signal handler for SIGALRM never gets removed. I
257 think that in my case it gets executed at some point later on when execution
258 has long left Curl_resolv_timeout() or even the cURL library.
260 The code that is jumped to with siglongjmp() simply sets the error message
261 to "name lookup timed out" and then returns with CURLRESOLV_ERROR. I guess
262 that instead of simply returning without cleaning up, the code should have a
263 goto that jumps to the spot right after the call to Curl_resolv().
265 Kamil Dudka (22 Mar 2010)
266 - Douglas Steinwand contributed a patch fixing insufficient initialization in
267 Curl_clone_ssl_config()
269 Daniel Stenberg (21 Mar 2010)
270 - Ben Greear improved TFTP: the error code returning and the treatment
271 of TSIZE == 0 when uploading.
273 - We've switched from CVS to git. See http://curl.haxx.se/source.html
275 Kamil Dudka (19 Mar 2010)
276 - Improved Curl_read() to not ignore the error returned from Curl_ssl_recv().
278 Daniel Stenberg (15 Mar 2010)
279 - Constantine Sapuntzakis brought a patch:
281 The problem mentioned on Dec 10 2009
282 (http://curl.haxx.se/bug/view.cgi?id=2905220) was only partially fixed.
283 Partially because an easy handle can be associated with many connections in
284 the cache (e.g. if there is a redirect during the lifetime of the easy
285 handle). The previous patch only cleaned up the first one. The new fix now
286 removes the easy handle from all connections, not just the first one.
288 Daniel Stenberg (6 Mar 2010)
289 - Ben Greear brought a patch that fixed the rate limiting logic for TFTP when
290 the easy interface was used.
292 Daniel Stenberg (5 Mar 2010)
293 - Daniel Johnson provided fixes for building curl with the clang compiler.
295 Yang Tse (5 Mar 2010)
296 - Constantine Sapuntzakis detected and fixed a double free in builds done
297 with threaded resolver enabled (Windows default configuration) that would
298 get triggered when a curl handle is closed while doing DNS resolution.
300 Daniel Stenberg (2 Mar 2010)
301 - [Daniel Johnson] I've been trying to build libcurl with clang on Darwin and
302 ran into some issues with the GSSAPI tests in configure.ac. The tests first
303 try to determine the include dirs and libs and set CPPFLAGS and LIBS
304 accordingly. It then checks for the headers and finally sets LIBS a second
305 time, causing the libs to be included twice. The first setting of LIBS seems
306 redundant and should be left out, since the first part is otherwise just
307 about finding headers.
309 My second issue is that 'krb5-config --libs gssapi' on Darwin is less than
310 useless and returns junk that, while it happens to work with gcc, causes
311 clang to choke. For example, --libs returns $CFLAGS along with the libs,
312 which is really retarded. Simply setting 'LIBS="$LIBS -lgssapi_krb5
313 -lresolv"' on Darwin is sufficient.
315 - Based on patch provided by Jacob Moshenko, the transfer logic now properly
316 makes sure that when using sub-second timeouts, there's no final bad 1000ms
317 wait. Previously, a sub-second timeout would often make the elapsed time end
318 up the time rounded up to the nearest second (e.g. 1s for 200ms timeout)
320 - Andrei Benea filed bug report #2956698 and pointed out that the
321 CURLOPT_CERTINFO feature leaked memory due to a missing OpenSSL function
322 call. He provided the patch to fix it too.
324 http://curl.haxx.se/bug/view.cgi?id=2956698
326 - Markus Duft pointed out in bug #2961796 that even though Interix has a
327 poll() function it doesn't quite work the way we want it so we must disable
328 it, and he also provided a patch for it.
330 http://curl.haxx.se/bug/view.cgi?id=2961796
332 - Made the pingpong timeout code properly deal with the response timeout AND
333 the global timeout if set. Also, as was reported in the bug report #2956437
334 by Ryan Chan, the time stamp to use as basis for the per command timeout was
335 not set properly in the DONE phase for FTP (and not for SMTP) so I fixed
336 that just now. This was a regression compared to 7.19.7 due to the
337 conversion of FTP code over to the generic pingpong concepts.
339 http://curl.haxx.se/bug/view.cgi?id=2956437
341 Daniel Stenberg (1 Mar 2010)
342 - Ben Greear provided an update for TFTP that fixes upload.
344 - Wesley Miaw reported bug #2958179 which identified a case of looping during
345 OpenSSL based SSL handshaking even though the multi interface was used and
346 there was no good reason for it.
348 http://curl.haxx.se/bug/view.cgi?id=2958179
350 Daniel Stenberg (26 Feb 2010)
351 - Pat Ray in bug #2958474 pointed out an off-by-one case when receiving a
352 chunked-encoding trailer.
354 http://curl.haxx.se/bug/view.cgi?id=2958474
356 Daniel Fandrich (25 Feb 2010)
357 - Fixed a couple of out of memory leaks and a segfault in the SMTP & IMAP code.
359 Yang Tse (25 Feb 2010)
360 - I fixed bug report #2958074 indicating
361 (http://curl.haxx.se/bug/view.cgi?id=2958074) that curl on Windows with
362 option --trace-time did not use local time when timestamping trace lines.
363 This could also happen on other systems depending on time souurce.
365 Patrick Monnerat (22 Feb 2010)
366 - Proper handling of STARTTLS on SMTP, taking CURLUSESSL_TRY into account.
367 - SMTP falls back to RFC821 HELO when EHLO fails (and SSL is not required).
368 - Use of true local host name (i.e.: via gethostname()) when available, as
369 default argument to SMTP HELO/EHLO.
370 - Test case 804 for HELO fallback.
372 Daniel Stenberg (20 Feb 2010)
373 - Fixed the SMTP compliance by making sure RCPT TO addresses are specified
374 properly in angle brackets. Recipients provided with CURLOPT_MAIL_RCPT now
375 get angle bracket wrapping automatically by libcurl unless the recipient
376 starts with an angle bracket as then the app is assumed to deal with that
379 - I made the SMTP code expect a 250 response back from the server after the
380 full DATA has been sent, and I modified the test SMTP server to also send
381 that response. As usual, the DONE operation that is made after a completed
382 transfer is still not doable in a non-blocking way so this waiting for 250
383 is unfortunately made blockingly.
385 Yang Tse (14 Feb 2010)
386 - Overhauled test suite getpart() function. Fixing potential out of bounds
387 stack and memory overwrites triggered with huge test case definitions.
389 Daniel Stenberg (13 Feb 2010)
390 - Martin Hager reported and fixed a problem with a missing quote in libcurl.m4
392 (http://curl.haxx.se/bug/view.cgi?id=2951319)
394 - Tom Donovan fixed the CURL_FORMAT_* defines when building with cmake.
396 (http://curl.haxx.se/bug/view.cgi?id=2951269)
398 Daniel Stenberg (12 Feb 2010)
399 - Jack Zhang reported a problem with SMTP: we wrongly used multiple addresses
400 in the same RCPT TO line, when they should be sent in separate single
401 commands. I updated test case 802 to verify this.
403 - I also fixed a bad use of my_setopt_str() of CURLOPT_MAIL_RCPT in the curl
404 tool which made it try to output it as string for the --libcurl feature
405 which could lead to crashes.
407 Yang Tse (11 Feb 2010)
408 - Steven M. Schweda fixed VMS builder bad behavior when used in a batch job,
409 removed obsolete batch_compile.com and defines.com and updated VMS readme.
411 Version 7.20.0 (9 February 2010)
413 Daniel Stenberg (9 Feb 2010)
414 - When downloading compressed content over HTTP and the app asked libcurl to
415 automatically uncompress it with the CURLOPT_ENCODING option, libcurl could
416 wrongly provide the callback with more data than the maximum documented
417 amount. An application could thus get tricked into badness if the maximum
418 limit was trusted to be enforced by libcurl itself (as it is documented).
420 This is further detailed and explained in the libcurl security advisory
423 http://curl.haxx.se/docs/adv_20100209.html
425 Daniel Fandrich (3 Feb 2010)
426 - Changed the Watcom makefiles to make them easier to keep in sync with
427 Makefile.inc since that can't be included directly.
429 Yang Tse (2 Feb 2010)
430 - Symbol CURL_FORMAT_OFF_T now obsoleted, will be removed in a future release,
431 symbol will not be available when building with CURL_NO_OLDIES defined. Use
432 of CURL_FORMAT_CURL_OFF_T is preferred since 7.19.0
434 Daniel Stenberg (1 Feb 2010)
435 - Using the multi_socket API, it turns out at times it seemed to "forget"
436 connections (which caused a hang). It turned out to be an existing (7.19.7)
437 bug in libcurl (that's been around for a long time) and it happened like
440 The app calls curl_multi_add_handle() to add a new easy handle, libcurl will
441 then set it to timeout in 1 millisecond so libcurl will tell the app about
444 The app's timeout fires off that there's a timeout, the app calls libcurl as
445 we so often document it:
448 res = curl_multi_socket_action(... TIMEOUT ...);
449 } while(CURLM_CALL_MULTI_PERFORM == res);
451 And this is the problem number one:
453 When curl_multi_socket_action() is called with no specific handle, but only
454 a timeout-action, it will *only* perform actions within libcurl that are
455 marked to run at this time. In this case, the request would go from INIT to
456 CONNECT and return CURLM_CALL_MULTI_PERFORM. When the app then calls libcurl
457 again, there's no timer set for this handle so it remains in the CONNECT
458 state. The CONNECT state is a transitional state in libcurl so it reports no
459 sockets there, and thus libcurl never tells the app anything more about that
460 easy handle/connection.
462 libcurl _does_ set a 1ms timeout for the handle at the end of
463 multi_runsingle() if it returns CURLM_CALL_MULTI_PERFORM, but since the loop
464 is instant the new job is not ready to run at that point (and there's no
465 code that makes libcurl call the app to update the timout for this new
466 timeout). It will simply rely on that some other timeout will trigger later
467 on or that something else will update the timeout callback. This makes the
468 bug fairly hard to repeat.
470 The fix made to adress this issue:
472 We introduce a loop in lib/multi.c around all calls to multi_runsingle() and
473 simply check for CURLM_CALL_MULTI_PERFORM internally. This has the added
474 benefit that this goes in line with my long-term wishes to get rid of the
475 CURLM_CALL_MULTI_PERFORM all together from the public API.
477 The downside of this fix, is that the counter we return in 'running_handles'
478 in several of our public functions then gets a slightly new and possibly
479 confusing behavior during times:
481 If an app adds a handle that fails to connect (very quickly) it may just
482 as well never appear as a 'running_handle' with this fix. Previously it
483 would first bump the counter only to get it decreased again at next call.
484 Even I have used that change in handle counter to signal "end of a
485 transfer". The only *good* way to find the end of a individual transfer
486 is calling curl_multi_info_read() to see if it returns one.
488 Of course, if the app previously did the looping before it checked the
489 counter, it really shouldn't be any new effect.
491 Yang Tse (26 Jan 2010)
492 - Constantine Sapuntzakis' and Joshua Kwan's work done in the last four months
493 relative to the asynchronous DNS lookups, along with with some integration
494 adjustments I have done are finally committed to CVS.
496 Currently these enhancements will benefit builds done using c-ares on any
497 platform as well as Windows builds using the default threaded resolver.
499 This release does not make generally available POSIX threaded DNS lookups
500 yet. There is no configure option to enable this feature yet. It is possible
501 to experimantally try this feature running configure with compiler flags that
502 make simultaneous definition of preprocessor symbols USE_THREADS_POSIX and
503 HAVE_PTHREAD_H, as well as whatever reentrancy compiler flags and linker ones
504 are required to link and properly use pthread_* functions on each platform.
506 Daniel Stenberg (26 Jan 2010)
507 - Mike Crowe made libcurl return CURLE_COULDNT_RESOLVE_PROXY when it is the
508 proxy that cannot be resolved when using c-ares. This matches the behaviour
509 when not using c-ares.
511 Björn Stenberg (23 Jan 2010)
512 - Added a new flag: -J/--remote-header-name. This option tells the
513 -O/--remote-name option to use the server-specified Content-Disposition
514 filename instead of extracting a filename from the URL.
516 Daniel Stenberg (21 Jan 2010)
517 - Chris Conroy brought support for RTSP transfers, and with it comes 8(!) new
518 libcurl options for controlling what to get and how to receive posssibly
519 interleaved RTP data.
521 Daniel Stenberg (20 Jan 2010)
522 - As was pointed out on the http-state mailing list, the order of cookies in a
523 HTTP Cookie: header _needs_ to be sorted on the path length in the cases
524 where two cookies using the same name are set more than once using
525 (overlapping) paths. Realizing this, identically named cookies must be
526 sorted correctly. But detecting only identically named cookies and take care
527 of them individually is harder than just to blindly and unconditionally sort
528 all cookies based on their path lengths. All major browsers also already do
529 this, so this makes our behavior one step closer to them in the cookie area.
531 Test case 8 was the only one that broke due to this change and I updated it
534 Daniel Stenberg (19 Jan 2010)
535 - David McCreedy brought a fix and a new test case (129) to make libcurl work
536 again when downloading files over FTP using ASCII and it turns out that the
537 final size of the file is not the same as the initial size the server
538 reported. This is very common since servers don't take the newline
539 conversions into account.
541 Kamil Dudka (14 Jan 2010)
542 - Suppressed side effect of OpenSSL configure checks, which prevented NSS from
543 being properly detected under certain circumstances. It had been caused by
544 strange behavior of pkg-config when handling PKG_CONFIG_LIBDIR. pkg-config
545 distinguishes among empty and non-existent environment variable in that case.
547 Daniel Stenberg (12 Jan 2010)
548 - Gil Weber reported a peculiar flaw with the multi interface when doing SFTP
549 transfers: curl_multi_fdset() would return -1 and not set and file
550 descriptors several times during a transfer of a single file. It turned out
551 to be due to two different flaws now fixed. Gil's excellent recipe helped me
554 Daniel Stenberg (11 Jan 2010)
555 - Made sure that the progress callback is repeatedly called at a regular
556 interval even during very slow connects.
558 - The tests/runtests.pl script now checks to see if the test case that runs is
559 present in the tests/data/Makefile.am and outputs a notice message on the
560 screen if not. Each test file has to be included in that Makefile.am to get
561 included in release archives and forgetting to add files there is a common
562 mistake. This is an attempt to make it harder to forget.
564 Daniel Stenberg (9 Jan 2010)
565 - Johan van Selst found and fixed a OpenSSL session ref count leak:
567 ossl_connect_step3() increments an SSL session handle reference counter on
568 each call. When sessions are re-used this reference counter may be
569 incremented many times, but it will be decremented only once when done (by
570 Curl_ossl_session_free()); and the internal OpenSSL data will not be freed
571 if this reference count remains positive. When a session is re-used the
572 reference counter should be corrected by explicitly calling
573 SSL_SESSION_free() after each consecutive SSL_get1_session() to avoid
574 introducing a memory leak.
576 (http://curl.haxx.se/bug/view.cgi?id=2926284)
578 Daniel Stenberg (7 Jan 2010)
579 - Make sure the progress callback is called repeatedly even during very slow
580 name resolves when c-ares is used for resolving.
582 Claes Jakobsson (6 Jan 2010)
583 - Julien Chaffraix fixed so that the fragment part in an URL is not sent
584 to the server anymore.
586 Kamil Dudka (3 Jan 2010)
587 - Julien Chaffraix eliminated a duplicated initialization in singlesocket().
589 Daniel Stenberg (2 Jan 2010)
590 - Make curl support --ssl and --ssl-reqd instead of the previous FTP-specific
591 versions --ftp-ssl and --ftp-ssl-reqd as these options are now used to
592 control SSL/TLS for IMAP, POP3 and SMTP as well in addition to FTP. The old
593 option names are still working but the new ones are the ones listed and
596 Daniel Stenberg (1 Jan 2010)
597 - Ingmar Runge enhanced libcurl's FTP engine to support the PRET command. This
598 command is a special "hack" used by the drftpd server, but even though it is
599 a custom extension I've deemed it fine to add to libcurl since this server
600 seems to survive and people keep using it and want libcurl to support
601 it. The new libcurl option is named CURLOPT_FTP_USE_PRET, and it is also
602 usable from the curl tool with --ftp-pret. Using this option on a server
603 that doesn't support this command will make libcurl fail.
605 I added test cases 1107 and 1108 to verify the functionality.
607 The PRET command is documented at
608 http://www.drftpd.org/index.php/Distributed_PASV
610 Yang Tse (30 Dec 2009)
611 - Steven M. Schweda improved VMS build system, and Craig A. Berry helped
612 with the patch and testing.
614 Daniel Stenberg (26 Dec 2009)
615 - Renato Botelho and Peter Pentchev brought a patch that makes the libcurl
616 headers work correctly even on FreeBSD systems before v8.
618 (http://curl.haxx.se/bug/view.cgi?id=2916915)
620 Daniel Stenberg (17 Dec 2009)
621 - David Byron fixed Curl_ossl_cleanup to actually call ENGINE_cleanup when
624 - Follow-up fix for the proxy fix I did for Jon Nelson's bug. It turned out I
625 was a bit too quick and broke test case 1101 with that change. The order of
626 some of the setups is sensitive. I now changed it slightly again to make
627 sure we do them in this order:
629 1 - parse URL and figure out what protocol is used in the URL
630 2 - prepend protocol:// to URL if missing
631 3 - parse name+password off URL, which needs to know what protocol is used
632 (since only some allows for name+password in the URL)
633 4 - figure out if a proxy should be used set by an option
634 5 - if no proxy option, check proxy environment variables
635 6 - run the protocol-specific setup function, which needs to have the proxy
638 Daniel Stenberg (15 Dec 2009)
639 - Jon Nelson found a regression that turned out to be a flaw in how libcurl
640 detects and uses proxies based on the environment variables. If the proxy
641 was given as an explicit option it worked, but due to the setup order
642 mistake proxies would not be used fine for a few protocols when picked up
643 from '[protocol]_proxy'. Obviously this broke after 7.19.4. I now also added
644 test case 1106 that verifies this functionality.
646 (http://curl.haxx.se/bug/view.cgi?id=2913886)
648 Daniel Stenberg (12 Dec 2009)
649 - IMAP, POP3 and SMTP support and their TLS versions (including IMAPS, POP3S
650 and SMTPS) are now supported. The current state may not yet be solid, but
651 the foundation is in place and the test suite has some initial support for
652 these protocols. Work will now persue to make them nice libcurl citizens
655 The work with supporting these new protocols was sponsored by
656 networking4all.com - thanks!
658 Daniel Stenberg (10 Dec 2009)
659 - Siegfried Gyuricsko found out that the curl manual said --retry would retry
660 on FTP errors in the transient 5xx range. Transient FTP errors are in the
661 4xx range. The code itself only tried on 5xx errors that occured _at login_.
662 Now the retry code retries on all FTP transfer failures that ended with a
665 (http://curl.haxx.se/bug/view.cgi?id=2911279)
667 - Constantine Sapuntzakis figured out a case which would lead to libcurl
668 accessing alredy freed memory and thus crash when using HTTPS (with
669 OpenSSL), multi interface and the CURLOPT_DEBUGFUNCTION and a certain order
670 of cleaning things up. I fixed it.
672 (http://curl.haxx.se/bug/view.cgi?id=2905220)
674 Daniel Stenberg (7 Dec 2009)
675 - Martin Storsjo made libcurl use the Expect: 100-continue header for posts
676 with unknown size. Previously it was only used for posts with a known size
677 larger than 1024 bytes.
679 Daniel Stenberg (1 Dec 2009)
680 - If the Expect: 100-continue header has been set by the application through
681 curl_easy_setopt with CURLOPT_HTTPHEADER, the library should set
682 data->state.expect100header accordingly - the current code (in 7.19.7 at
683 least) doesn't handle this properly. Martin Storsjo provided the fix!
685 Yang Tse (28 Nov 2009)
686 - Added Diffie-Hellman parameters to several test harness certificate files in
687 PEM format. Required by several stunnel versions used by our test harness.
689 Daniel Stenberg (28 Nov 2009)
690 - Markus Koetter provided a polished and updated version of Chad Monroe's TFTP
691 rework patch that now integrates TFTP properly into libcurl so that it can
692 be used non-blocking with the multi interface and more. BLKSIZE also works.
694 The --tftp-blksize option was added to allow setting the TFTP BLKSIZE from
697 Daniel Stenberg (26 Nov 2009)
698 - Extended and fixed the change I did on Dec 11 for the the progress
699 meter/callback during FTP command/response sequences. It turned out it was
700 really lame before and now the progress meter SHOULD get called at least
703 Daniel Stenberg (23 Nov 2009)
704 - Bjorn Augustsson reported a bug which made curl not report any problems even
705 though it failed to write a very small download to disk (done in a single
706 fwrite call). It turned out to be because fwrite() returned success, but
707 there was insufficient error-checking for the fclose() call which tricked
708 curl to believe things were fine.
710 Yang Tse (23 Nov 2009)
711 - David Byron modified Makefile.dist vc8 and vc9 targets in order to allow
712 finer granularity control when generating src and lib makefiles.
714 Yang Tse (22 Nov 2009)
715 - I modified configure to force removal of the curlbuild.h file included in
716 distribution tarballs for use by non-configure systems. As intended, this
717 would get overwriten when doing in-tree builds. But VPATH builds would end
718 having two curlbuild.h files, one in the source tree and another in the
719 build tree. With the modification I introduced 5 Nov 2009 this could become
720 an issue when running libcurl's test suite.
722 Daniel Stenberg (20 Nov 2009)
723 - Constantine Sapuntzakis identified a write after close, as the sockets were
724 closed by libcurl before the SSL lib were shutdown and they may write to its
725 socket. Detected to at least happen with OpenSSL builds.
727 - Jad Chamcham pointed out a bug with connection re-use. If a connection had
728 CURLOPT_HTTPPROXYTUNNEL enabled over a proxy, a subsequent request using the
729 same proxy with the tunnel option disabled would still wrongly re-use that
730 previous connection and the outcome would only be badness.
732 Yang Tse (18 Nov 2009)
733 - I modified the memory tracking system to make it intolerant with zero sized
734 malloc(), calloc() and realloc() function calls.
736 Daniel Stenberg (17 Nov 2009)
737 - Constantine Sapuntzakis provided another fix for the DNS cache that could
738 end up with entries that wouldn't time-out:
740 1. Set up a first web server that redirects (307) to a http://server:port
742 2. Have curl connect to the first web server using curl multi
744 After the curl_easy_cleanup call, there will be curl dns entries hanging
745 around with in_use != 0.
747 (http://curl.haxx.se/bug/view.cgi?id=2891591)
749 - Marc Kleine-Budde fixed: curl saved the LDFLAGS set during configure into
750 its pkg-config file. So -Wl stuff ended up in the .pc file, which is really
751 bad, and breaks if there are multiple -Wl in our LDFLAGS (which are in
752 PTXdist). bug #2893592 (http://curl.haxx.se/bug/view.cgi?id=2893592)
754 Kamil Dudka (15 Nov 2009)
755 - David Byron improved the configure script to use pkg-config to find OpenSSL
756 (and in particular the list of required libraries) even if a path is given
757 as argument to --with-ssl
759 Yang Tse (15 Nov 2009)
760 - I removed enable-thread / disable-thread configure option. These were only
761 placebo options. The library is always built as thread safe as possible on
764 Claes Jakobsson (14 Nov 2009)
765 - curl-config now accepts '--configure' to see what arguments was
766 passed to the configure script when building curl.
768 Daniel Stenberg (14 Nov 2009)
769 - Claes Jakobsson restored the configure functionality to detect NSS when
770 --with-nss is set but not "yes".
772 I think we can still improve that to check for pkg-config in that path etc,
773 but at least this patch brings back the same functionality we had before.
775 - Camille Moncelier added support for the file type SSL_FILETYPE_ENGINE for
776 the client certificate. It also disable the key name test as some engines
777 can select a private key/cert automatically (When there is only one key
778 and/or certificate on the hardware device used by the engine)
780 Yang Tse (14 Nov 2009)
781 - Constantine Sapuntzakis provided the fix that ensures that an SSL connection
782 won't be reused unless protection level for peer and host verification match.
784 I refactored how preprocessor symbol _THREAD_SAFE definition is done.
786 Kamil Dudka (12 Nov 2009)
787 - Kevin Baughman provided a fix preventing libcurl-NSS from crash on doubly
788 closed NSPR descriptor. The issue was hard to find, reported several times
789 before and always closed unresolved. More info at the RH bug:
790 https://bugzilla.redhat.com/534176
792 - libcurl-NSS now tries to reconnect with TLS disabled in case it detects
793 a broken TLS server. However it does not happen if SSL version is selected
794 manually. The approach was originally taken from PSM. Kaspar Brand helped me
795 to complete the patch. Original bug reports:
796 https://bugzilla.redhat.com/525496
797 https://bugzilla.redhat.com/527771
799 Yang Tse (12 Nov 2009)
800 - I modified configure script to make the getaddrinfo function check also
801 verify if the function is thread safe.
803 Yang Tse (11 Nov 2009)
804 - Marco Maggi reported that compilation failed when configured --with-gssapi
805 and GNU GSS installed due to a missing mutual exclusion of header files in
806 the Kerberos 5 code path. He also verified that my patch worked for him.
808 Daniel Stenberg (11 Nov 2009)
809 - Constantine Sapuntzakis posted bug #2891595
810 (http://curl.haxx.se/bug/view.cgi?id=2891595) which identified how an entry
811 in the DNS cache would linger too long if the request that added it was in
812 use that long. He also provided the patch that now makes libcurl capable of
813 still doing a request while the DNS hash entry may get timed out.
815 - Christian Schmitz noticed that the progress meter/callback was not properly
816 used during the FTP connection phase (after the actual TCP connect), while
817 it of course should be. I also made the speed check get called correctly so
818 that really slow servers will trigger that properly too.
820 Kamil Dudka (5 Nov 2009)
821 - Dropped misleading timeouts in libcurl-NSS and made sure the SSL socket works
822 in non-blocking mode.
824 Yang Tse (5 Nov 2009)
825 - I removed leading 'curl' path on the 'curlbuild.h' include statement in
826 curl.h, adjusting auto-makefiles include path, to enhance portability to
827 OS's without an orthogonal directory tree structure such as OS/400.
829 Daniel Stenberg (4 Nov 2009)
830 - I fixed several problems with the transfer progress meter. It showed the
831 wrong percentage for small files, most notable for <1000 bytes and could
832 easily end up showing more than 100% at the end. It also didn't show any
833 percentage, transfer size or estimated transfer times when transferring
836 Version 7.19.7 (4 November 2009)
838 Daniel Stenberg (2 Nov 2009)
839 - As reported independent by both Stan van de Burgt and Didier Brisebourg,
840 CURLINFO_SIZE_DOWNLOAD (the -w variable size_download) didn't work when
841 getting data from ldap!
843 Daniel Stenberg (31 Oct 2009)
844 - Gabriel Kuri reported a problem with CURLINFO_CONTENT_LENGTH_DOWNLOAD if the
845 download was 0 bytes, as libcurl would then return the size as unknown (-1)
846 and not 0. I wrote a fix and test case 566 to verify it.
848 Daniel Stenberg (30 Oct 2009)
849 - Liza Alenchery mentioned a problem with re-used SCP connection when a bad
850 auth is used, as it caused a crash. I failed to repeat the issue, but still
851 made a change that now forces the TCP connection used for a freed SCP
852 session to get closed and not be re-used.
854 - "Tom" posted a bug report that mentioned how libcurl did wrong when doing a
855 POST using a read callback, with Digest authentication and
856 "Transfer-Encoding: chunked" enforced. I would then cause the first request
857 to be wrongly sent and then basically hang until the server closed the
858 connection. I fixed the problem and added test case 565 to verify it.
860 Daniel Stenberg (25 Oct 2009)
861 - Dima Barsky made the curl cookie parser accept cookies even with blank or
862 unparsable expiry dates and then treat them as session cookies - previously
863 libcurl would reject cookies with a date format it couldn't parse. Research
864 shows that the major browser treat such cookies as session cookies. I
865 modified test 8 and 31 to verify this.
867 Daniel Stenberg (21 Oct 2009)
868 - Attempt to use pkg-config for finding out libssh2 installation details
871 - A patch in bug report #2883177 (http://curl.haxx.se/bug/view.cgi?id=2883177)
872 by Johan van Selst introduced the --crlfile option to curl, which makes curl
873 tell libcurl about a file with CRL (certificate revocation list) data to
876 Daniel Stenberg (18 Oct 2009)
877 - Ray Dassen provided a patch in Debian's bug tracker (bug number #551461)
878 that now makes curl_getdate(3) actually handles RFC 822 formatted dates that
879 use the "single letter military timezones".
880 http://www.rfc-ref.org/RFC-TEXTS/822/chapter5.html has the details.
882 - Fixed memory leak in the SCP/SFTP code as it never freed the knownhosts
885 - John Dennis filed bug report #2873666
886 (http://curl.haxx.se/bug/view.cgi?id=2873666) which identified a problem
887 which made libcurl loop infinitely when given incorrect credentials when
888 using HTTP GSS negotiate authentication. He also provided a small and simple
891 - Kevin Baughman found a double close() problem with libcurl-NSS, as when
892 libcurl called NSS to close the SSL "session" it also closed the actual
895 Yang Tse (17 Oct 2009)
896 - Bug report #2866724 indicated
897 (http://curl.haxx.se/bug/view.cgi?id=2866724) that curl on Windows failed
898 when writing files whose file names originally contained characters which
899 are not valid for file names on Windows. Dan Fandrich provided an initial
900 patch and another revised one to fix this issue.
902 Daniel Stenberg (1 Oct 2009)
903 - Tom Mueller correctly reported in bug report #2870221
904 (http://curl.haxx.se/bug/view.cgi?id=2870221) that libcurl returned an
905 incorrect return code from the internal trynextip() function which caused
906 him grief. This is a regression that was introduced in 7.19.1 and I find it
907 strange it hasn't hit us harder, but I won't persue into figuring out
910 - Constantine Sapuntzakis: The current implementation will always set
911 SO_SNDBUF to CURL_WRITE_SIZE even if the SO_SNDBUF starts out larger. The
912 patch doesn't do a setsockopt if SO_SNDBUF is already greater than
913 CURL_WRITE_SIZE. This should help folks who have set up their computer with
916 Daniel Stenberg (27 Sep 2009)
917 - I introduced a maximum limit for received HTTP headers. It is controlled by
918 the define CURL_MAX_HTTP_HEADER which is even exposed in the public header
919 file to allow for users to fairly easy rebuild libcurl with a modified
920 limit. The rationale for a fixed limit is that libcurl is realloc()ing a
921 buffer to be able to put a full header into it, so that it can call the
922 header callback with the entire header, but that also risk getting it into
923 trouble if a server by mistake or willingly sends a header that is more or
924 less without an end. The limit is set to 100K.
926 Daniel Stenberg (26 Sep 2009)
927 - John P. McCaskey posted a bug report that showed how libcurl did wrong when
928 saving received cookies with no given path, if the path in the request had a
929 query part. That is means a question mark (?) and characters on the right
930 side of that. I wrote test case 1105 and fixed this problem.
932 Kamil Dudka (26 Sep 2009)
933 - Implemented a protocol independent way to specify blocking direction, used by
934 transfer.c for blocking. It is currently used only by SCP and SFTP protocols.
935 This enhancement resolves an issue with 100% CPU usage during SFTP upload,
938 Daniel Stenberg (25 Sep 2009)
939 - Chris Mumford filed bug report #2861587
940 (http://curl.haxx.se/bug/view.cgi?id=2861587) identifying that libcurl used
941 the OpenSSL function X509_load_crl_file() wrongly and failed if it would
942 load a CRL file with more than one certificate within. This is now fixed.
944 Daniel Stenberg (16 Sep 2009)
945 - Sven Anders reported that we introduced a cert verfication flaw for OpenSSL-
946 powered libcurl in 7.19.6. If there was a X509v3 Subject Alternative Name
947 field in the certficate it had to match and so even if non-DNS and non-IP
948 entry was present it caused the verification to fail.
950 Daniel Fandrich (15 Sep 2009)
951 - Moved the libssh2 checks after the SSL library checks. This helps when
952 statically linking since libssh2 needs the SSL library link flags to be
953 set up already to satisfy its dependencies. This wouldn't be necessary if
954 the libssh2 configure check was changed to use pkg-config since the
955 --static flag would add the dependencies automatically.
957 Yang Tse (14 Sep 2009)
958 - Revert Joshua Kwan's patch committed 11 Sep 2009.
960 Some systems poll function sets POLLHUP in revents without setting
961 POLLIN, and sets POLLERR without setting POLLIN and POLLOUT. In some
962 libcurl code execution paths this could trigger busy wait loops with
963 high CPU usage until a timeout condition aborted the loop.
965 The reverted patch addressed the above issue for a very specific case,
966 when awaiting c-ares to resolve. A libcurl-wide fix for Curl_poll now
969 Guenter Knauf (11 Sep 2009)
970 - Joshua Kwan provided a patch to pass POLLERR / POLLHUP back to c-ares.
971 This fixes a loop problem with high CPU usage.
973 Daniel Stenberg (10 Sep 2009)
974 - Claes Jakobsson fixed a problem with cookie expiry dates at exctly the epoch
975 start second "Thu Jan 1 00:00:00 GMT 1970" as the date parser then returns 0
976 which internally then is treated as a session cookie. That particular date
977 is now made to get the value of 1.
979 Daniel Stenberg (2 Sep 2009)
980 - Daniel Johnson found a flaw in the code converting sftp-errors to libcurl
983 Daniel Stenberg (1 Sep 2009)
984 - Peter Sylvester made a debug feature for Curl_resolv() that now will force
985 libcurl to resolve 'localhost' whatever name you use in the URL *if* you set
986 the --interface option to (exactly) "LocalHost". This will enable us to
987 write tests for custom hosts names but still use a local host server.
989 - configure now tries to use pkg-config for a number of sub-dependencies even
990 when cross-compiling. The key to success is then you properly setup
991 PKG_CONFIG_PATH before invoking configure.
993 I also improved how NSS is detected by trying nss-config if pkg-config isn't
994 present, and as a last resort just use the lib name and force the user to
995 setup the LIBS/LDFLAGS/CFLAGS etc properly. The previous last resort would
996 add a range of various libs that would almost never be quite correct.
998 Daniel Stenberg (31 Aug 2009)
999 - When using the multi interface with FTP and you asked for NOBODY, you did no
1000 QUOTE commands and the request used the same path as the connection had
1001 already changed to, it would decide that no commands would be necessary for
1002 the "DO" action and that was not handled properly but libcurl would instead
1005 Kamil Dudka (28 Aug 2009)
1006 - Improved error message for not matching certificate subject name in
1007 libcurl-NSS. Originally reported at:
1008 https://bugzilla.redhat.com/show_bug.cgi?id=516056#c9
1010 Patrick Monnerat (24 Aug 2009)
1011 - Introduced a SYST-based test to properly set-up name format when dealing
1012 with the OS/400 FTP server.
1014 - Fixed an ftp_readresp() bug preventing detection of failing control socket
1015 and causing FTP client to loop forever.
1017 Daniel Stenberg (24 Aug 2009)
1018 - Marc de Bruin pointed out that configure --with-gnutls=PATH didn't work
1019 properly and provided a fix. http://curl.haxx.se/bug/view.cgi?id=2843008
1021 - Eric Wong introduced support for the new option -T. (dot) that makes curl
1022 read stdin in a non-blocking fashion. This also brings back -T- (minus) to
1023 the previous blocking behavior since it could break stuff for people at
1026 Michal Marek (21 Aug 2009)
1027 - With CURLOPT_PROXY_TRANSFER_MODE, avoid sending invalid URLs like
1028 ftp://example.com;type=i if the user specified ftp://example.com without the
1031 Daniel Stenberg (21 Aug 2009)
1032 - Andre Guibert de Bruet pointed out a missing return code check for a
1033 strdup() that could lead to segfault if it returned NULL. I extended his
1034 suggest patch to now have Curl_retry_request() return a regular return code
1035 and better check that.
1037 - Lots of good work by Krister Johansen, mostly related to pipelining:
1039 Fix SIGSEGV on free'd easy_conn when pipe unexpectedly breaks
1040 Fix data corruption issue with re-connected transfers
1041 Fix use after free if we're completed but easy_conn not NULL
1043 Kamil Dudka (13 Aug 2009)
1044 - Changed NSS code to not ignore the value of ssl.verifyhost and produce more
1045 verbose error messages. Originally reported at:
1046 https://bugzilla.redhat.com/show_bug.cgi?id=516056
1048 Daniel Stenberg (12 Aug 2009)
1049 - Karl Moerder fixed the Makefile.vc* makefiles to include the new file
1050 nonblock.c so that they work fine again
1052 - I expanded test 517 with a bunch of more dates that originate from the
1053 Chrome browser test suite. It turns out most of them get parsed the same
1056 Version 7.19.6 (12 August 2009)
1058 Daniel Stenberg (12 Aug 2009)
1059 - Carsten Lange reported a bug and provided a patch for TFTP upload and the
1060 sending of the TSIZE option. I don't like fixing bugs just hours before
1061 a release, but since it was broken and the patch fixes this for him I decided
1062 to get it in anyway.
1064 Daniel Stenberg (11 Aug 2009)
1065 - Peter Sylvester made the HTTPS test server use specific certificates for
1066 each test, so that the test suite can now be used to actually test the
1067 verification of cert names etc. This made an error show up in the OpenSSL-
1068 specific code where it would attempt to match the CN field even if a
1069 subjectAltName exists that doesn't match. This is now fixed and verified
1072 - Benbuck Nason posted the bug report #2835196
1073 (http://curl.haxx.se/bug/view.cgi?id=2835196), fixing a few compiler
1074 warnings when mixing ints and bools.
1076 Daniel Fandrich (10 Aug 2009)
1077 - Fixed a memory leak in the FTP code and an off-by-one heap buffer overflow.
1079 Daniel Fandrich (9 Aug 2009)
1080 - Fixed some memory leaks in the command-line tool that caused most of the
1081 torture tests to fail.
1083 Daniel Stenberg (2 Aug 2009)
1084 - Curt Bogmine reported a problem with SNI enabled on a particular server. We
1085 should introduce an option to disable SNI, but as we're in feature freeze
1086 now I've addressed the obvious bug here (pointed out by Peter Sylvester): we
1087 shouldn't try to enable SNI when SSLv2 or SSLv3 is explicitly selected.
1088 Code for OpenSSL and GnuTLS was fixed. NSS doesn't seem to have a particular
1089 option for SNI, or are we simply not using it?
1091 Daniel Stenberg (1 Aug 2009)
1092 - Scott Cantor posted the bug report #2829955
1093 (http://curl.haxx.se/bug/view.cgi?id=2829955) mentioning the recent SSL cert
1094 verification flaw found and exploited by Moxie Marlinspike. The presentation
1095 he did at Black Hat is available here:
1096 https://www.blackhat.com/html/bh-usa-09/bh-usa-09-archives.html#Marlinspike
1098 Apparently at least one CA allowed a subjectAltName or CN that contain a
1099 zero byte, and thus clients that assumed they would never have zero bytes
1100 were exploited to OK a certificate that didn't actually match the site. Like
1101 if the name in the cert was "example.com\0theatualsite.com", libcurl would
1102 happily verify that cert for example.com.
1104 libcurl now better uses the length of the extracted name, not using the zero
1105 termination for getting the string length.
1107 This fixing only made and needed in OpenSSL interfacing code.
1109 - Tanguy Fautre pointed out that OpenSSL's function RAND_screen() (present
1110 only in some OpenSSL installs - like on Windows) isn't thread-safe and we
1111 agreed that moving it to the global_init() function is a decent way to deal
1112 with this situation.
1114 - Alexander Beedie provided the patch for a noproxy problem: If I have set
1115 CURLOPT_NOPROXY to "*", or to a host that should not use a proxy, I actually
1116 could still end up using a proxy if a proxy environment variable was set.
1118 Daniel Stenberg (27 Jul 2009)
1119 - All the quote options (CURLOPT_QUOTE, CURLOPT_POSTQUOTE and
1120 CURLOPT_PREQUOTE) now accept a preceeding asterisk before the command to
1121 send when using FTP, as a sign that libcurl shall simply ignore the response
1122 from the server instead of treating it as an error. Not treating a 400+ FTP
1123 response code as an error means that failed commands will not abort the
1124 chain of commands, nor will they cause the connection to get disconnected.
1126 Daniel Stenberg (26 Jul 2009)
1127 - Johan van Selst posted bug report #2825989
1128 (http://curl.haxx.se/bug/view.cgi?id=2825989) pointing out that
1129 OpenSSL-powered libcurl didn't support the SHA-2 digest algorithm, and
1130 provided the solution too: to use OpenSSL_add_all_algorithms() in addition
1131 to the older SSLeay_* alternative. OpenSSL_add_all_algorithms was added in
1134 Daniel Stenberg (23 Jul 2009)
1135 - Added CURLOPT_SSH_KNOWNHOSTS, CURLOPT_SSH_KEYFUNCTION, CURLOPT_SSH_KEYDATA.
1136 They introduce known_host support for SSH keys to libcurl. See docs for
1137 details. Note that this feature depends on a new enough libssh2 version, to
1138 be supported in libssh2 1.2 and later (or current git repo at this time).
1140 Michal Marek (22 Jul 2009)
1141 - David Binderman found a memory and fd leak in lib/gtls.c:load_file()
1142 (https://bugzilla.novell.com/523919). When looking at the code, I found that
1143 also the ptr pointer can leak.
1145 Kamil Dudka (20 Jul 2009)
1146 - Claes Jakobsson improved the support for client certificates handling in
1147 NSS-powered libcurl. Now the client certificates can be selected
1148 automatically by a NSS built-in hook. Additionally pre-login to all PKCS11
1149 slots is no more performed. It used to cause problems with HW tokens.
1151 - Fixed reference counting for NSS client certificates. Now the PEM reader
1152 module should be always properly unloaded on Curl_nss_cleanup(). If the
1153 unload fails though, libcurl will try to reuse the already loaded instance.
1155 Daniel Fandrich (15 Jul 2009)
1156 - Added nonblock.c to the non-automake makefiles (note that the dependencies
1157 in the Watcom makefiles aren't quite correct).
1159 Michal Marek (15 Jul 2009)
1160 - Changed the description of CURLINFO_OS_ERRNO to make it clear that the
1161 errno is not reset on success.
1163 Guenter Knauf (14 Jul 2009)
1164 - renamed generated config.h to curl_config.h to avoid any future clashes
1165 with config.h from other projects.
1167 Daniel Stenberg (9 Jul 2009)
1168 - Eric Wong introduced curlx_nonblock() that the curl tool now (re-)uses for
1169 setting a file descriptor non-blocking. Used by the functionality Eric
1170 himself brough on June 15th.
1172 Daniel Stenberg (8 Jul 2009)
1173 - Constantine Sapuntzakis posted bug report #2813123
1174 (http://curl.haxx.se/bug/view.cgi?id=2813123) and an a patch that fixes the
1177 Url A is accessed using auth. Url A redirects to Url B (on a different
1178 server0. Url B reuses a persistent connection. Url B has auth, even though
1179 it's on a different server.
1181 Note: if Url B does not reuse a persistent connection, auth is not sent.
1185 data->state.first_host is not initialized becuase Curl_http_connect is not
1186 called when a connection is reused.
1190 move initialization of data->state.first_host to Curl_http. No code before
1191 Curl_http uses data->state.first_host anyway.
1193 Guenter Knauf (4 Jul 2009)
1194 - Markus Koetter provided a patch to avoid getnameinfo() usage which broke a
1195 couple of both IPv4 and IPv6 autobuilds.
1197 Daniel Stenberg (29 Jun 2009)
1198 - Markus Koetter made CURLOPT_FTPPORT (and curl's -P/--ftpport) support a port
1199 range if given colon-separated after the host name/address part. Like
1200 "192.168.0.1:2000-10000"
1202 - Modified the separators used for CURLOPT_CERTINFO in multi-part outputs. I
1203 don't know how they got wrong in the first place, but using this output
1204 format makes it possible to quite easily separate the string into an array
1207 Daniel Fandrich (16 June 2009)
1208 - Added a few more compiler warning options for gcc.
1210 Daniel Stenberg (16 Jun 2009)
1211 - Reuven Wachtfogel made curl -o - properly produce a binary output on windows
1212 (no newline translations). Use -B/--use-ascii if you rather get the ascii
1215 Michal Marek (16 Jun 2009)
1216 - When doing non-anonymous ftp via http proxies and the password is not
1217 provided in the url, add it there (squid needs this).
1219 Daniel Stenberg (15 Jun 2009)
1220 - Eric Wong's patch:
1222 This allows curl(1) to be used as a client-side tunnel for arbitrary stream
1223 protocols by abusing chunked transfer encoding in both the HTTP request and
1224 HTTP response. This requires server support for sending a response while a
1225 request is still being read, of course.
1227 If attempting to read from stdin returns EAGAIN, then we pause our sender.
1228 This leaves curl to attempt to read from the socket while reading from stdin
1229 (and thus sending) is paused.
1231 This change was needed to allow successfully tunneling the git protocol over
1232 HTTP (--no-buffer is needed, as well).
1234 Patrick Monnerat (15 Jun 2009)
1235 - Replaced use of standard C library rand()/srand() by our own pseudo-random
1238 Yang Tse (11 Jun 2009)
1239 - I adapted testcurl script to allow building test harness programs when
1240 cross-compiling for a *-*-mingw* host.
1242 Daniel Stenberg (10 Jun 2009)
1243 - Fabian Keil ran clang on the (lib)curl code, found a bunch of warnings and
1244 contributed a range of patches to fix them.
1246 Yang Tse (10 Jun 2009)
1247 - I introduced configure script option --enable-curldebug which now allows
1248 the decoupled enabling or disabling of the curl debug memory tracking
1249 feature from the --enable-debug option which no longer controls this.
1251 curl --version will list 'Debug' feature for debug enabled builds, and
1252 will list 'TrackMemory' feature for curl debug memory tracking capable
1253 builds. These features are independent and can be controlled when running
1254 the configure script. When --enable-debug is given both features will be
1255 enabled, unless some restriction prevents memory tracking from being used.
1257 Internally, definition of preprocessor symbol DEBUGBUILD restricts code
1258 which is only compiled for debug enabled builds. And symbol CURLDEBUG is
1259 used to differentiate code which is _only_ used for memory tracking.
1261 Yang Tse (9 Jun 2009)
1262 - Daniel Steinberg pointed out that Curl_FormInit() in formdata.c was not
1263 initializing the fread callback pointer and this triggered a compiler
1264 warning, also provided a friendly suggestion on how to fix it.
1266 Daniel Stenberg (8 Jun 2009)
1267 - Claes Jakobsson provided a patch for libcurl-NSS that fixed a bad refcount
1268 issue with client certs that caused issues like segfaults.
1269 http://curl.haxx.se/mail/lib-2009-05/0316.html
1271 - Triggered by bug report #2798852 and the patch in there, I fixed configure
1272 to detect gnutls build options with pkg-config only and not libgnutls-config
1273 anymore since GnuTLS has stopped distributing that tool. If an explicit path
1274 is given to configure, we will instead guess on how to link and use that
1275 lib. I did not use the patch from the bug report.
1277 Yang Tse (8 Jun 2009)
1278 - Igor Novoseltsev adjusted Makefile.vxworks to get sources and headers
1279 included from Makefile.inc, and provided docs\INSTALL VxWorks section.
1281 - I removed buildconf.bat from release and daily snapshot archives. This
1282 file is only for CVS tree checkout builds.
1284 Daniel Stenberg (8 Jun 2009)
1285 - Eric Wong fixed --no-buffer to actually switch off output buffering. Been
1288 Bill Hoffman (6 Jun 2009)
1289 - Added some cmake docs and fixed socklen_t in the build.
1291 Yang Tse (5 Jun 2009)
1292 - John E. Malmberg provided VMS specific patch: "This fixes an existing bug
1293 in urlglob.c where it was not converting the Curl Unix exit code to a VMS
1294 DCL compatible exit code. This fix required the enhancement described next.
1295 This also adds an enhancement to main.c so that when curl is run under a
1296 Unix shell like Bash on VMS, it will return the standard Unix exit codes
1297 and messages." And another patch for docs/examples.
1299 I introduced os-specific.c and os-specific.h for use in curl tool code
1300 and adjusted John E. Malmberg's patch placement to use these new files
1301 as an effort to prevent main.c from growing ad infinitum. Code already
1302 existing in main.c which is OS specific should be moved into these files.
1304 Daniel Stenberg (4 June 2009)
1305 - Setting the Content-Length: header from your app when you do a POST or PUT
1306 is almost always a VERY BAD IDEA. Yet there are still apps out there doing
1307 this, and now recently it triggered a bug/side-effect in libcurl as when
1308 libcurl sends a POST or PUT with NTLM, it sends an empty post first when it
1309 knows it will just get a 401/407 back. If the app then replaced the
1310 Content-Length header, it caused the server to wait for input that libcurl
1311 wouldn't send. Aaron Oneal reported this problem in bug report #2799008
1312 (http://curl.haxx.se/bug/view.cgi?id=2799008) and helped us verify the fix.
1314 Yang Tse (4 Jun 2009)
1315 - Igor Novoseltsev provided patches and information, that after some
1316 adjustments to better fit curl's way of doing things, have resulted
1317 in the posibility of building libcurl for VxWorks.
1319 Daniel Fandrich (2 June 2009)
1320 - Checked in a Google Android make file. To use it, you must first
1321 create a config.h file by running configure in the Android environment,
1322 which doesn't seem to be easy to do. If no easy way can be found, a
1323 static config-android.h may need to be created and checked in to the
1324 libcurl source tree.
1326 Daniel Stenberg (1 June 2009)
1327 - Claes Jakobsson fixed the configure script to better find and use NSS
1330 Yang Tse (1 Jun 2009)
1331 - John E. Malmberg provided a VMS specific clean-up for curl.h, and pointed
1332 out that the configure script was failing to detect the timeval struct on
1333 VMS when building with _XOPEN_SOURCE_EXTENDED undefined due to definition
1334 taking place in socket.h instead of time.h. I have adjusted configure
1335 script to also include this header when checking struct timeval.
1337 Daniel Stenberg (27 May 2009)
1338 - Frank McGeough provided a small OpenSSL #include fix to make libcurl compile
1339 fine with Nokia 5th edition 1.0 SDK for Symbian.
1341 - Andre Guibert de Bruet found a call to a OpenSSL function that didn't check
1342 for a failure properly.
1344 - Mike Crowe pointed out that setting CURLOPT_USERPWD to NULL used to clear
1345 the auth credentials back in 7.19.0 and earlier while now you have to set ""
1346 to get the same effect. His patch brings back the ability to use NULL.
1348 - Claes Jakobsson fixed libcurl-NSS to build fine even without the
1349 PK11_CreateGenericObject() function.
1351 Daniel Stenberg (25 May 2009)
1352 - bug report #2796358 (http://curl.haxx.se/bug/view.cgi?id=2796358) pointed
1353 out that the cookie parser would leak memory when it parses cookies that are
1354 received with domain, path etc set multiple times in the same header. While
1355 such a cookie is questionable, they occur in the wild and libcurl no longer
1356 leaks memory for them. I added such a header to test case 8.
1358 Daniel Fandrich (22 May 2009)
1359 - Removed some obsolete digest code that caused a valgrind error in test 551.
1361 Daniel Fandrich (20 May 2009)
1362 - Added "non-existing host" test keywords to make it easy to skip those
1363 tests on machines that have broken DNS configurations (such as
1364 those configured to use OpenDNS).
1366 Daniel Stenberg (19 May 2009)
1367 - Kamil Dudka brought the patch from the Redhat bug entry
1368 https://bugzilla.redhat.com/show_bug.cgi?id=427966 which was libcurl closing
1369 a bad file descriptor when closing down the FTP data connection. Caolan
1370 McNamara seems to be the original author of it.
1372 Version 7.19.5 (18 May 2009)
1374 Daniel Stenberg (17 May 2009)
1375 - James Bursa posted a patch to the mailing list that fixed a problem with
1376 no_proxy which made it not skip the proxy if the URL entered contained a
1377 user name. I added test case 1101 to verify.
1379 Daniel Stenberg (11 May 2009)
1380 - Balint Szilakszi reported a memory leak when libcurl did gzip decompression
1381 of streams that had some parts (legitimately) missing. We now provide and use
1382 a proper cleanup function for the content encoding submodule.
1383 http://curl.haxx.se/mail/lib-2009-05/0092.html
1385 - Kamil Dudka provided a fix for libcurl-NSS reported by Michael Cronenworth
1386 at https://bugzilla.redhat.com/show_bug.cgi?id=453612#c12
1388 If an incorrect password is given while loading a private key, libcurl ends
1389 up in an infinite loop consuming memory. The bug is critical.
1391 - I fixed the problem with doing NTLM, POST and then following a 302 redirect,
1392 as reported by Ebenezer Ikonne (on curl-users) and Laurent Rabret (on
1393 curl-library). The transfer was mistakenly marked to get more data to send
1394 but since it didn't actually have that, it just hung there...
1396 Daniel Stenberg (10 May 2009)
1397 - Andre Guibert de Bruet correctly pointed out an over-alloc with one wasted
1398 byte in the digest code.
1400 Yang Tse (9 May 2009)
1401 - Removed DOS and TPF package's subdirectory Makefile.am, it was only used
1402 to include some files in the distribution tarball serving no other purpose.
1403 Files from the DOS and TPF subdirectories are now included in the EXTRA_DIST
1404 of the Makefile in the parent subdirectory.
1406 Yang Tse (8 May 2009)
1407 - Changed host name literal in several tests to one under the haxx.se domain.
1409 - Renamed vc6 workspace and project files to avoid filename clash when used
1410 for conversion to later VS versions.
1412 Daniel Stenberg (8 May 2009)
1413 - Constantine Sapuntzakis fixed bug report #2784055
1414 (http://curl.haxx.se/bug/view.cgi?id=2784055) identifying a problem to
1415 connect to SOCKS proxies when using the multi interface. It turned out to
1416 almost not work at all previously. We need to wait for the TCP connect to
1417 be properly verified before doing the SOCKS magic.
1419 There's still a flaw in the FTP code for this.
1421 Daniel Stenberg (7 May 2009)
1422 - Made the SO_SNDBUF setting for the data connection socket for ftp uploads as
1423 well. See change 28 Apr 2009.
1425 Yang Tse (7 May 2009)
1426 - Fixed an issue affecting FTP transfers, introduced with the transfer.c
1427 patch committed May 4.
1429 Daniel Stenberg (7 May 2009)
1430 - Man page *roff problems fixed thanks to input from Colin Watson. Problems
1431 reported in the Debian package.
1433 - Vijay G filed bug report #2723236
1434 (http://curl.haxx.se/bug/view.cgi?id=2723236) identifying a problem with
1435 libcurl's TFTP code and its lack of dealing with the OACK packet.
1437 Yang Tse (5 May 2009)
1438 - Fixed the --ftp-port address of test #251 to the CLIENTIP address, and
1439 reverted the change affecting test suite harness committed 4 May.
1441 Daniel Stenberg (5 May 2009)
1442 - Inspired by Michael Smith's session id fix for OpenSSL, I did the
1443 corresponding fix in the GnuTLS code: make sure to store the new session id
1444 in case the previous re-used one is rejected.
1446 Daniel Stenberg (4 May 2009)
1447 - Michael Smith posted bug report #2786255
1448 (http://curl.haxx.se/bug/view.cgi?id=2786255) with a patch, identifying how
1449 libcurl did not deal with SSL session ids properly if the server rejected a
1450 re-use of one. Starting now, it will forget the rejected one and remember
1451 the new. This change was for OpenSSL only, it is likely that other SSL lib
1452 code needs similar fixes.
1454 Yang Tse (4 May 2009)
1455 - Applied David McCreedy's "transfer.c fixes for CURL_DO_LINEEND_CONV and
1456 non-ASCII platform HTTP requests" patch addressing two HTTP PUT problems:
1457 1) On non-ASCII platforms not all of the protocol portions of the PUT are
1458 being translated to ASCII. 2) On all platforms the line endings of part of
1459 the protocol portions are mangled from CRLF to CRCRLF if data->set.crlf or
1460 data->set.prefer_ascii are set (depending on CURL_DO_LINEEND_CONV).
1462 - Applied David McCreedy's patch to fix test suite harness to allow test FTP
1463 server and client on different machines, providing FTP client address when
1464 running the FTP test server.
1466 Daniel Fandrich (3 May 2009)
1467 - Added and disabled test case 563 which shows KNOWN_BUGS #59. The bug
1468 report failed to mention that a proxy must be used to reproduce it.
1470 Yang Tse (2 May 2009)
1471 - Use a build-time configured curl_socklen_t data type instead of socklen_t.
1473 Yang Tse (1 May 2009)
1474 - Applied David McCreedy's patches "TPF-platform specific changes to various
1475 files" and "http.c fix to Curl_proxyCONNECT for non-ASCII platforms", the
1476 former with minor edits.
1478 Daniel Stenberg (30 Apr 2009)
1479 - I was going to fix issue #59 in KNOWN_BUGS
1481 If the CURLOPT_PORT option is used on an FTP URL like
1482 "ftp://example.com/file;type=A" the ";type=A" is stripped off.
1484 I added test case 562 to verify, only to find out that I couldn't repeat
1485 this bug so I hereby consider it not a bug anymore!
1487 Daniel Stenberg (29 Apr 2009)
1488 - Based on bug report #2723219 (http://curl.haxx.se/bug/view.cgi?id=2723219)
1489 I've now made TFTP "connections" not being kept for re-use within libcurl.
1490 TFTP is UDP-based so the benefit was really low (if even existing) to begin
1491 with so instead of tracking down to fix this problem we instead removed the
1492 re-use. I also enabled test case 1099 that I wrote a few days ago to verify
1493 that this change fixes the reported problem.
1495 Daniel Stenberg (28 Apr 2009)
1496 - Constantine Sapuntzakis filed bug report #2783090
1497 (http://curl.haxx.se/bug/view.cgi?id=2783090) pointing out that on windows
1498 we need to grow the SO_SNDBUF buffer somewhat to get really good upload
1499 speeds. http://support.microsoft.com/kb/823764 has the details. Friends
1500 confirmed that simply adding 32 to CURL_MAX_WRITE_SIZE is enough.
1502 - Bug report #2709004 (http://curl.haxx.se/bug/view.cgi?id=2709004) by Tim
1503 Chen pointed out how curl couldn't upload with resume when reading from a
1506 This ended up with the introduction of a new return code for the
1507 CURLOPT_SEEKFUNCTION callback that basically says that the seek failed but
1508 that libcurl may try to resolve the situation anyway. In our case this means
1509 libcurl will attempt to instead read that much data from the stream instead
1510 of seeking and that way curl can now upload with resume when data is read
1513 Daniel Stenberg (26 Apr 2009)
1514 - Bug report #2779733 (http://curl.haxx.se/bug/view.cgi?id=2779733) by Sven
1515 Wegener pointed out that CURLINFO_APPCONNECT_TIME didn't work with the multi
1516 interface and provided a patch that fixed the problem!
1518 Daniel Stenberg (24 Apr 2009)
1519 - Kamil Dudka fixed another NSS-related leak when client certs were used.
1521 - Bug report #2779245 (http://curl.haxx.se/bug/view.cgi?id=2779245) by Rainer
1522 Koenig pointed out that the man page didn't tell that the *_proxy
1523 environment variables can be specified lower case or UPPER CASE and the
1524 lower case takes precedence,
1526 Daniel Fandrich (21 Apr 2009)
1527 - Added new libcurl source files to Amiga, RiscOS and VC6 build files.
1529 Yang Tse (21 Apr 2009)
1530 - Moved potential inclusion of system's malloc.h and memory.h header files to
1531 setup_once.h. Inclusion of each header file is based on the definition of
1532 NEED_MALLOC_H and NEED_MEMORY_H respectively.
1534 Renamed libcurl's memory.h to curl_memory.h
1536 Daniel Stenberg (20 Apr 2009)
1537 - Leanic Lefever reported a crash and did some detailed research on why and
1538 how it occurs (http://curl.haxx.se/mail/lib-2009-04/0289.html). The
1539 conclusion was that if an error is detected and Curl_done() is called for
1540 the connection, ftp_done() could at times return another error code that
1541 then would take precedence and that new code confused existing logic that
1542 works for the first error code (CURLE_SEND_ERROR) only.
1544 - Gisle Vanem noticed that --libtool would produce bogus strings at times for
1545 OBJECTPOINT options. Now we've introduced a new function - my_setopt_str -
1546 within the app for setting plain string options to avoid the risk of this
1549 Daniel Stenberg (17 Apr 2009)
1550 - Pramod Sharma reported and tracked down a bug when doing FTP over a HTTP
1551 proxy. libcurl would then wrongly close the connection after each
1552 request. In his case it had the weird side-effect that it killed NTLM auth
1553 for the proxy causing an inifinite loop!
1555 I added test case 1098 to verify this fix. The test case does however not
1556 properly verify that the transfers are done persistently - as I couldn't
1557 think of a clever way to achieve it right now - but you need to read the
1558 stderr output after a test run to see that it truly did the right thing.
1560 Daniel Stenberg (13 Apr 2009)
1561 - bug report #2727981 (http://curl.haxx.se/bug/view.cgi?id=2727981) by Martin
1562 Storsjö pointed out how setting CURLOPT_NOBODY to 0 could be downright
1563 confusing as it set the method to either GET or HEAD. The example he showed
1566 curl_easy_setopt(curl, CURLOPT_PUT, 1);
1567 curl_easy_setopt(curl, CURLOPT_NOBODY, 0);
1569 The new way doesn't alter the method until the request is about to start. If
1570 CURLOPT_NOBODY is then 1 the HTTP request will be HEAD. If CURLOPT_NOBODY is
1571 0 and the request happens to have been set to HEAD, it will then instead be
1572 set to GET. I believe this will be less surprising to users, and hopefully
1573 not hit any existing users badly.
1575 - Toshio Kuratomi reported a memory leak problem with libcurl+NSS that turned
1576 out to be leaking cacerts. Kamil Dudka helped me complete the fix. The issue
1577 is found in Redhat's bug tracker:
1578 https://bugzilla.redhat.com/show_bug.cgi?id=453612
1580 There are still memory leaks present, but they seem to have other reasons.
1582 Daniel Fandrich (11 Apr 2009)
1583 - Added new libcurl source files to Symbian OS build files.
1584 - Improved Symbian support for SSL.
1586 Yang Tse (10 Apr 2009)
1587 - Daniel Johnson improved the MacOSX-Framework shell script to now perform all
1588 the steps required to build a Mac OS X four way fat ppc/i386/ppc64/x86_64
1589 libcurl.framework. Four way fat framework requires OS X 10.5 SDK or later.
1591 Yang Tse (8 Apr 2009)
1592 - Removed Sun compilers preprocessor block from curlbuild.h.dist, this also
1593 removes it from the curlbuild.h file originally distributed by the cURL
1594 project as this file is intended for systems not capable of running the
1595 configure script. For those who have been building curl out of the source
1596 code curl distribution tarball provided by curl.haxx.se the change implies
1597 nothing. Previous change in this area committed 2 Apr becomes irrelevant.
1599 Daniel Stenberg (6 Apr 2009)
1600 - I clarified in the docs that CURLOPT_SEEKFUNCTION should return 0 on success
1601 and 1 on fatal errors. Previously it only mentioned non-zero on fatal
1602 errors. This is a slight change in meaning, but it follows what we've done
1603 elsewhere before and it opens up for LOTS of more useful return codes
1604 whenever we can think of them...
1606 Yang Tse (2 Apr 2009)
1607 - Fix curl_off_t definition for builds done using Sun compilers and a
1608 non-configured libcurl. In this case curl_off_t data type was gated
1609 to the off_t data type which depends on the _FILE_OFFSET_BITS. This
1610 configuration is exactly the unwanted configuration for our curl_off_t
1611 data type which must not depend on such setting. This breaks ABI for
1612 libcurl libraries built with Sun compilers which were built without
1613 having run the configure script with _FILE_OFFSET_BITS different than
1614 64 and using the ILP32 data model.
1616 Daniel Stenberg (1 Apr 2009)
1617 - Andre Guibert de Bruet fixed a NULL pointer use in an infof() call if a
1618 strdup() call failed.
1620 Daniel Fandrich (31 Mar 2009)
1621 - Properly return an error code in curl_easy_recv (reported by Jim Freeman).
1623 Daniel Stenberg (18 Mar 2009)
1624 - Kamil Dudka brought a patch that enables 6 additional crypto algorithms when
1625 NSS is used. These ciphers were added in NSS 3.4 and require to be enabled
1628 Daniel Stenberg (13 Mar 2009)
1629 - Use libssh2_version() to present the libssh2 version in case the libssh2
1630 library is found to support it.
1632 Yang Tse (12 Mar 2009)
1633 - Added missing Curl_read() return code checking in TELNET transfers.
1635 - Pierre Brico found and fixed TELNET transfers not being aborted upon
1636 a write callback failure.
1638 Daniel Stenberg (11 Mar 2009)
1639 - Kamil Dudka made the curl tool properly call curl_global_init() before any
1640 other libcurl function.
1642 Yang Tse (11 Mar 2009)
1643 - Added missing TELNET timeout support for Windows builds. This issue was
1644 reported by Pierre Brico.
1646 Daniel Stenberg (9 Mar 2009)
1647 - Frank Hempel found out a bug and provided the fix:
1649 curl_easy_duphandle did not necessarily duplicate the CURLOPT_COOKIEFILE
1650 option. It only enabled the cookie engine in the destination handle if
1651 data->cookies is not NULL (where data is the source handle). In case of a
1652 newly initialized handle which just had the cookie support enabled by a
1653 curl_easy_setopt(handle, CURL_COOKIEFILE, "")-call, handle->cookies was
1654 still NULL because the setopt-call only appends the value to
1655 data->change.cookielist, hence duplicating this handle would not have the
1656 cookie engine switched on.
1658 We also concluded that the slist-functionality would be suitable for being
1659 put in its own module rather than simply hanging out in lib/sendf.c so I
1660 created lib/slist.[ch] for them.
1662 - Andreas Farber made the 'buildconf' script check for the presence of m4
1663 scripts to make it detect a bad checkout earlier. People with older
1664 checkouts who don't do cvs update with the -d option won't get the new dirs
1665 and then will get funny outputs that can be a bit hard to understand and
1668 Daniel Stenberg (8 Mar 2009)
1669 - Andre Guibert de Bruet found and fixed a code segment in ssluse.c where the
1670 allocation of the memory BIO was not being properly checked.
1672 - Andre Guibert de Bruet fixed the gnutls-using code: There are a few places
1673 in the gnutls code where we were checking for negative values for errors,
1674 when the man pages state that GNUTLS_E_SUCCESS is returned on success and
1675 other values indicate error conditions.
1677 - Bill Egert pointed out (http://curl.haxx.se/bug/view.cgi?id=2671602) that
1678 curl didn't use sprintf() in a way that is documented to work in POSIX but
1679 since we use our own printf() code (from libcurl) that shouldn't be a
1680 problem. Nonetheless I modified the code to not rely on such particular
1681 features and to not cause further raised eyebrowse with no good reason.
1683 Daniel Fandrich (5 Mar 2009)
1684 - Expanded the security section of the libcurl-tutorial man page to cover
1685 more issues for authors to consider when writing robust libcurl-using
1688 Yang Tse (5 Mar 2009)
1689 - Fixed NTLM authentication memory leak on SSPI enabled Windows builds. This
1690 issue was noticed by Chris Deidun.
1692 Daniel Fandrich (4 Mar 2009)
1693 - Fixed a problem with m4 quoting in the OpenSSL configure check reported
1696 Daniel Stenberg (3 Mar 2009)
1697 - David James brought a patch that make libcurl close (all) dead connections
1698 whenever you attempt to open a new connection.
1700 1. After cleaning up a dead connection, "continue" instead of
1701 returning FALSE. This ensures that we clean up all dead connections,
1702 rather than just cleaning up the first dead connection.
1703 2. Move up the cleanup for dead connections so that it occurs for
1704 all connections, rather than just the connections which have the same
1705 preferences as our current new connection.
1707 Version 7.19.4 (3 March 2009)
1709 Daniel Stenberg (3 Mar 2009)
1710 - David Kierznowski notified us about a security flaw
1711 (http://curl.haxx.se/docs/adv_20090303.html also known as CVE-2009-0037) in
1712 which previous libcurl versions (by design) can be tricked to access an
1713 arbitrary local/different file instead of a remote one when
1714 CURLOPT_FOLLOWLOCATION is enabled. This flaw is now fixed in this release
1715 together this the addition of two new setopt options for controlling this
1718 o CURLOPT_REDIR_PROTOCOLS controls what protocols libcurl is allowed to
1719 follow to when CURLOPT_FOLLOWLOCATION is enabled. By default, this option
1720 excludes the FILE and SCP protocols and thus you nee to explicitly allow
1721 them in your app if you really want that behavior.
1723 o CURLOPT_PROTOCOLS controls what protocol(s) libcurl is allowed to fetch
1724 using the primary URL option. This is useful if you want to allow a user or
1725 other outsiders control what URL to pass to libcurl and yet not allow all
1726 protocols libcurl may have been built to support.
1728 Daniel Stenberg (27 Feb 2009)
1729 - Senthil Raja Velu reported a problem when CURLOPT_INTERFACE and
1730 CURLOPT_LOCALPORT were used together (the local port bind failed), and
1731 Markus Koetter provided the fix!
1733 Daniel Stenberg (25 Feb 2009)
1734 - As Daniel Fandrich figured out, we must do the GnuTLS initing in the
1735 curl_global_init() function to properly maintain the performing functions
1736 thread-safe. We've previously (28 April 2007) moved the init to a later time
1737 just to avoid it to fail very early when libgcrypt dislikes the situation,
1738 but that move was bad and the fix should rather be in libgcrypt or
1741 Daniel Stenberg (24 Feb 2009)
1742 - Brian J. Murrell found out that Negotiate proxy authentication didn't work.
1743 It happened because the code used the struct for server-based auth all the
1744 time for both proxy and server auth which of course was wrong.
1746 Daniel Stenberg (23 Feb 2009)
1747 - After a bug reported by James Cheng I've made curl_easy_getinfo() for
1748 CURLINFO_CONTENT_LENGTH_DOWNLOAD and CURLINFO_CONTENT_LENGTH_UPLOAD return
1749 -1 if the sizes aren't know. Previously these returned 0, make it impossible
1750 to detect the difference between actually zero and unknown.
1752 Yang Tse (23 Feb 2009)
1753 - Daniel Johnson provided a shell script that will perform all the steps needed
1754 to build a Mac OS X fat ppc/i386 or ppc64/x86_64 libcurl.framework
1756 Daniel Stenberg (23 Feb 2009)
1757 - I renamed everything in the windows builds files that used the name 'curllib'
1758 to the proper 'libcurl' as clearly this caused confusion.
1760 Yang Tse (20 Feb 2009)
1761 - Do not halt compilation when using VS2008 to build a Windows 2000 target.
1763 Daniel Stenberg (20 Feb 2009)
1764 - Linus Nielsen Feltzing reported and helped me repeat and fix a problem with
1765 FTP with the multi interface: when a transfer fails, like when aborted by a
1766 write callback, the control connection was wrongly closed and thus not
1769 This change is also an attempt to cleanup the code somewhat in this area, as
1770 now the FTP code attempts to keep (better) track on pending responses
1771 necessary to get read in ftp_done().
1773 Daniel Stenberg (19 Feb 2009)
1774 - Patrik Thunstrom reported a problem and helped me repeat it. It turned out
1775 libcurl did a superfluous 1000ms wait when doing SFTP downloads!
1777 We read data with libssh2 while doing the "DO" operation for SFTP and then
1778 when we were about to start getting data for the actual file part, the
1779 "TRANSFER" part, we waited for socket action (in 1000ms) before doing a
1780 libssh2-read. But in this case libssh2 had already read and buffered the
1781 data so we ended up always just waiting 1000ms before we get working on the
1784 Patrick Monnerat (18 Feb 2009)
1785 - FTP downloads (i.e.: RETR) ending with code 550 now return error
1786 CURLE_REMOTE_FILE_NOT_FOUND instead of CURLE_FTP_COULDNT_RETR_FILE.
1788 Daniel Stenberg (17 Feb 2009)
1789 - Kamil Dudka made NSS-powered builds compile and run again!
1791 - A second follow-up change by Andre Guibert de Bruet to fix a related memory
1792 leak like that fixed on the 14th. When zlib returns failure, we need to
1793 cleanup properly before returning error.
1795 - CURLOPT_FTP_CREATE_MISSING_DIRS can now be set to 2 in addition to 1 for
1796 plain FTP connections, and it will then allow MKD to fail once and retry the
1797 CWD afterwards. This is especially useful if you're doing many simultanoes
1798 connections against the same server and they all have this option enabled,
1799 as then CWD may first fail but then another connection does MKD before this
1800 connection and thus MKD fails but trying CWD works! The numbers can
1801 (should?) now be set with the convenience enums now called
1802 CURLFTP_CREATE_DIR and CURLFTP_CREATE_DIR_RETRY.
1804 Tests has proven that if you're making an application that uploads a set of
1805 files to an ftp server, you will get a noticable gain in speed if you're
1806 using multiple connections and this option will be then be very useful.
1808 Daniel Stenberg (14 Feb 2009)
1809 - Andre Guibert de Bruet found and fixed a memory leak in the content encoding
1810 code, which could happen on libz errors.
1812 Daniel Fandrich (12 Feb 2009)
1813 - Added support for Digest and NTLM authentication using GnuTLS.
1815 Daniel Stenberg (11 Feb 2009)
1816 - CURLINFO_CONDITION_UNMET was added to allow an application to get to know if
1817 the condition in the previous request was unmet. This is typically a time
1818 condition set with CURLOPT_TIMECONDITION and was previously not possible to
1819 reliably figure out. From bug report #2565128
1820 (http://curl.haxx.se/bug/view.cgi?id=2565128) filed by Jocelyn Jaubert.
1822 Daniel Fandrich (4 Feb 2009)
1823 - Don't add the standard /usr/lib or /usr/include paths to LDFLAGS and CPPFLAGS
1824 (respectively) when --with-ssl=/usr is used (patch based on FreeBSD).
1826 - Added an explicit buffer limit check in msdosify() (patch based on FreeBSD).
1827 This couldn't ever overflow in curl, but might if the code were used
1828 elsewhere or under different conditions.
1830 Daniel Stenberg (3 Feb 2009)
1831 - Hidemoto Nakada provided a small fix that makes it possible to get the
1832 CURLINFO_CONTENT_LENGTH_DOWNLOAD size from file:// "transfers" with
1833 CURLOPT_NOBODY set true.
1835 Daniel Stenberg (2 Feb 2009)
1836 - Patrick Scott found a rather large memory leak when using the multi
1837 interface and setting CURLMOPT_MAXCONNECTS to something less than the number
1838 of handles you add to the multi handle. All the connections that didn't fit
1839 in the cache would not be properly disconnected nor freed!
1841 - Craig A West brought us: libcurl now defaults to do CONNECT with HTTP
1842 version 1.1 instead of 1.0 like before. This change also introduces the new
1843 proxy type for libcurl called 'CURLPROXY_HTTP_1_0' that then allows apps to
1844 switch (back) to CONNECT 1.0 requests. The curl tool also got a --proxy1.0
1845 option that works exactly like --proxy but sets CURLPROXY_HTTP_1_0.
1847 I updated all test cases cases that use CONNECT and I tried to do some using
1848 --proxy1.0 and some updated to do CONNECT 1.1 to get both versions run.
1850 Daniel Stenberg (31 Jan 2009)
1851 - When building with c-ares 1.6.1 (not yet released) or later and IPv6 support
1852 enabled, we can now take advantage of its brand new AF_UNSPEC support in
1853 ares_gethostbyname(). This makes test case 241 finally run fine for me with
1854 this setup since it now parses the "::1 ip6-localhost" line fine in my
1857 Daniel Stenberg (30 Jan 2009)
1858 - Scott Cantor filed bug report #2550061
1859 (http://curl.haxx.se/bug/view.cgi?id=2550061) mentioning that I failed to
1860 properly make sure that the VC9 makefiles got included in the latest
1861 release. I've now fixed the release script and verified it so next release
1862 will hopefully include them properly!
1864 Daniel Fandrich (30 Jan 2009)
1865 - Fixed --disable-proxy for FTP and SOCKS. Thanks to Daniel Egger for
1868 Yang Tse (29 Jan 2009)
1869 - Introduced curl_sspi.c and curl_sspi.h for the implementation of functions
1870 Curl_sspi_global_init() and Curl_sspi_global_cleanup() which previously were
1871 named Curl_ntlm_global_init() and Curl_ntlm_global_cleanup() in http_ntlm.c
1872 Also adjusted socks_sspi.c to remove the link-time dependency on the Windows
1873 SSPI library using it now in the same way as it was done in http_ntlm.c.
1875 Daniel Stenberg (28 Jan 2009)
1876 - Markus Moeller introduced two new options to libcurl:
1877 CURLOPT_SOCKS5_GSSAPI_SERVICE and CURLOPT_SOCKS5_GSSAPI_NEC to allow libcurl
1878 to do GSS-style authentication with SOCKS5 proxies. The curl tool got the
1879 options called --socks5-gssapi-service and --socks5-gssapi-nec to enable
1882 Daniel Stenberg (26 Jan 2009)
1883 - Chad Monroe provided the new CURLOPT_TFTP_BLKSIZE option that allows an app
1884 to set desired block size to use for TFTP transfers instead of the default
1887 - The "-no_ticket" option was introduced in Openssl0.9.8j. It's a flag to
1888 disable "rfc4507bis session ticket support". rfc4507bis was later turned
1889 into the proper RFC5077 it seems: http://tools.ietf.org/html/rfc5077
1891 The enabled extension concerns the session management. I wonder how often
1892 libcurl stops a connection and then resumes a TLS session. also, sending the
1893 session data is some overhead. .I suggest that you just use your proposed
1894 patch (which explicitly disables TICKET).
1896 If someone writes an application with libcurl and openssl who wants to
1897 enable the feature, one can do this in the SSL callback.
1899 Sharad Gupta brought this to my attention. Peter Sylvester helped me decide
1900 on the proper action.
1902 - Alexey Borzov filed bug report #2535504
1903 (http://curl.haxx.se/bug/view.cgi?id=2535504) pointing out that realms with
1904 quoted quotation marks in HTTP Digest headers didn't work. I've now added
1905 test case 1095 that verifies my fix.
1907 - Craig A West brought CURLOPT_NOPROXY and the corresponding --noproxy option.
1908 They basically offer the same thing the NO_PROXY environment variable only
1909 offered previously: list a set of host names that shall not use the proxy
1910 even if one is specified.
1912 Daniel Fandrich (20 Jan 2009)
1913 - Call setlocale() for libtest tests to test the effects of locale-induced
1914 libc changes on libcurl.
1916 - Fixed a couple more locale-dependent toupper conversions, mainly for
1917 clarity. This does fix one problem that causes ;type=i FTP URLs
1918 to fail in the Turkish locale when CURLOPT_PROXY_TRANSFER_MODE is
1919 used (test case 561)
1921 - Added tests 561 and 1091 through 1094 to test various combinations
1922 of ;type= and ;mode= URLs that could potentially fail in the Turkish
1925 Daniel Stenberg (20 Jan 2009)
1926 - Lisa Xu pointed out that the ssh.obj file was missing from the
1927 lib/Makefile.vc6 file (and thus from the vc8 and vc9 ones too).
1929 Version 7.19.3 (19 January 2009)
1931 Daniel Stenberg (16 Jan 2009)
1932 - Andrew de los Reyes fixed curlbuild.h for "generic" gcc builds on PPC, both
1935 Daniel Stenberg (15 Jan 2009)
1936 - Tim Ansell fixed a compiler warning in lib/cookie.c
1938 Daniel Stenberg (14 Jan 2009)
1939 - Grant Erickson fixed timeouts for TFTP such that specifying a
1940 connect-timeout, a max-time or both options work correctly and as expected
1941 by passing the correct boolean value to Curl_timeleft via the
1942 'duringconnect' parameter.
1944 With this small change, curl TFTP now behaves as expected (and likely as
1945 originally-designed):
1947 1) For non-existent or unreachable dotted IP addresses:
1949 a) With no options, follows the default curl 300s timeout...
1950 b) With --connect-timeout only, follows that value...
1951 c) With --max-time only, follows that value...
1952 d) With both --connect-timeout and --max-time, follows the smaller value...
1954 and times out with a "curl: (7) Couldn't connect to server" error.
1956 2) For transfers to/from a valid host:
1958 a) With no options, follows default curl 300s timeout for the
1959 first XRQ/DATA/ACK transaction and the default TFTP 3600s
1960 timeout for the remainder of the transfer...
1962 b) With --connect-time only, follows that value for the
1963 first XRQ/DATA/ACK transaction and the default TFTP 3600s
1964 timeout for the remainder of the transfer...
1966 c) With --max-time only, follows that value for the first
1967 XRQ/DATA/ACK transaction and for the remainder of the
1970 d) With both --connect-timeout and --max-time, follows the former
1971 for the first XRQ/DATA/ACK transaction and the latter for the
1972 remainder of the transfer...
1974 and times out with a "curl: (28) Timeout was reached" error as
1977 Daniel Stenberg (13 Jan 2009)
1978 - Michael Wallner fixed a NULL pointer deref when calling
1979 curl_easy_setup(curl, CURLOPT_COOKIELIST, "SESS") on a CURL handle with no
1982 - Stefan Teleman brought a patch to fix the default curlbuild.h file for the
1985 Daniel Stenberg (12 Jan 2009)
1986 - Based on bug report #2498665 (http://curl.haxx.se/bug/view.cgi?id=2498665)
1987 by Daniel Black, I've now added magic to the configure script that makes it
1988 use pkg-config to detect gnutls details as well if the existing method
1989 (using libgnutls-config) fails. While doing this, I cleaned up and unified
1990 the pkg-config usage when detecting openssl and nss as well.
1992 Daniel Stenberg (11 Jan 2009)
1993 - Karl Moerder brought the patch that creates vc9 Makefiles, and I made
1994 'maketgz' now use the actual makefile targets to do the VC8 and VC9
1997 Daniel Stenberg (10 Jan 2009)
1998 - Emil Romanus fixed:
2000 When using the multi interface over HTTP and the server returns a Location
2001 header, the running easy handle will get stuck in the CURLM_STATE_PERFORM
2002 state, leaving the external event loop stuck waiting for data from the
2003 ingoing socket (when using the curl_multi_socket_action stuff). While this
2004 bug was pretty hard to find, it seems to require only a one-line fix. The
2005 break statement on line 1374 in multi.c caused the function to skip the call
2008 How to reproduce this bug? Well, that's another question. evhiperfifo.c in
2009 the examples directory chokes on this bug only _sometimes_, probably
2010 depending on how fast the URLs are added. One way of testing the bug out is
2011 writing to hiper.fifo from more than one source at the same time.
2013 Daniel Fandrich (7 Jan 2009)
2014 - Unified much of the SessionHandle initialization done in Curl_open() and
2015 curl_easy_reset() by creating Curl_init_userdefined(). This had the side
2016 effect of fixing curl_easy_reset() so it now also resets
2017 CURLOPT_FTP_FILEMETHOD and CURLOPT_SSL_SESSIONID_CACHE
2019 Daniel Stenberg (7 Jan 2009)
2020 - Rob Crittenden did once again provide an NSS update:
2022 I have to jump through a few hoops now with the NSS library initialization
2023 since another part of an application may have already initialized NSS by the
2024 time Curl gets invoked. This patch is more careful to only shutdown the NSS
2025 library if Curl did the initialization.
2027 It also adds in a bit of code to set the default ciphers if the app that
2028 call NSS_Init* did not call NSS_SetDomesticPolicy() or set specific
2029 ciphers. One might argue that this lets other application developers get
2030 lazy and/or they aren't using the NSS API correctly, and you'd be right.
2031 But still, this will avoid terribly difficult-to-trace crashes and is
2034 Daniel Stenberg (1 Jan 2009)
2035 - 'reconf' is removed since we rather have users use 'buildconf'
2037 Daniel Stenberg (31 Dec 2008)
2038 - Bas Mevissen reported http://curl.haxx.se/bug/view.cgi?id=2479030 pointing
2039 out that 'reconf' didn't properly point out the m4 subdirectory when running
2042 Daniel Stenberg (29 Dec 2008)
2043 - Phil Lisiecki filed bug report #2413067
2044 (http://curl.haxx.se/bug/view.cgi?id=2413067) that identified a problem that
2045 would cause libcurl to mark a DNS cache entry "in use" eternally if the
2046 subsequence TCP connect failed. It would thus never get pruned and refreshed
2047 as it should've been.
2049 Phil provided his own patch to this problem that while it seemed to work
2050 wasn't complete and thus I wrote my own fix to the problem.
2052 Daniel Stenberg (28 Dec 2008)
2053 - Peter Korsgaard fixed building libcurl with "configure --with-ssl
2056 - Anthony Bryan fixed more language and spelling flaws in man pages.
2058 Daniel Stenberg (22 Dec 2008)
2059 - Given a recent enough libssh2, libcurl can now seek/resume with SFTP even
2060 on file indexes beyond 2 or 4GB.
2062 - Anthony Bryan provided a set of patches that cleaned up manual language,
2063 corrected spellings and more.
2065 Daniel Stenberg (20 Dec 2008)
2066 - Igor Novoseltsev fixed a bad situation for the multi_socket() API when doing
2067 pipelining, as libcurl could then easily get confused and A) work on the
2068 handle that was not "first in queue" on a pipeline, or even B) tell the app
2069 to REMOVE a socket while it was in use by a second handle in a pipeline. Both
2070 errors caused hanging or stalling applications.
2072 Daniel Stenberg (19 Dec 2008)
2073 - curl_multi_timeout() could return a timeout value of 0 even though nothing
2074 was actually ready to get done, as the internal time resolution is higher
2075 than the returned millisecond timer. Therefore it could cause applications
2076 running on fast processors to do short bursts of busy-loops.
2077 curl_multi_timeout() will now only return 0 if the timeout is actually
2080 - Using the libssh2 0.19 function libssh2_session_block_directions(), libcurl
2081 now has an improved ability to do right when the multi interface (both
2082 "regular" and multi_socket) is used for SCP and SFTP transfers. This should
2083 result in (much) less busy-loop situations and thus less CPU usage with no
2086 Daniel Stenberg (17 Dec 2008)
2087 - SCP and SFTP with the multi interface had the same flaw: the 'DONE'
2088 operation didn't complete properly if the EAGAIN equivalent was returned but
2089 libcurl would simply continue with a half-completed close operation
2090 performed. This ruined persistent connection re-use and cause some
2091 SSH-protocol errors in general. The correction is unfortunately adding a
2092 blocking function - doing it entirely non-blocking should be considered for
2095 Gisle Vanem (16 Dec 2008)
2096 - Added the possibility to use the Watt-32 tcp/ip stack under Windows.
2097 The change simply involved adding a USE_WATT32 section in the
2098 config-win32.h files (under ./lib and ./src). This section disables
2099 the use of any Winsock headers.
2101 Daniel Stenberg (16 Dec 2008)
2102 - libssh2_sftp_last_error() was wrongly used at some places in libcurl which
2103 made libcurl sometimes not properly abort problematic SFTP transfers.
2105 Daniel Stenberg (12 Dec 2008)
2106 - More work with Igor Novoseltsev to first fix the remaining stuff for
2107 removing easy handles from multi handles when the easy handle is/was within
2108 a HTTP pipeline. His bug report #2351653
2109 (http://curl.haxx.se/bug/view.cgi?id=2351653) was also related and was
2110 eventually fixed by a patch by Igor himself.
2112 Yang Tse (12 Dec 2008)
2113 - Patrick Monnerat fixed a build regression, introduced in 7.19.2, affecting
2114 OS/400 compilations with IPv6 enabled.
2116 Daniel Stenberg (12 Dec 2008)
2117 - Mark Karpeles filed bug report #2416182 titled "crash in ConnectionExists
2118 when using duphandle+curl_mutli"
2119 (http://curl.haxx.se/bug/view.cgi?id=2416182) which showed that
2120 curl_easy_duphandle() wrongly also copied the pointer to the connection
2121 cache, which was plain wrong and caused a segfault if the handle would be
2122 used in a different multi handle than the handle it was duplicated from.
2124 Daniel Stenberg (11 Dec 2008)
2125 - Keshav Krity found out that libcurl failed to deal with dotted IPv6
2126 addresses if they were very long (>39 letters) due to a too strict address
2127 validity parser. It now accepts addresses up to 45 bytes long.
2129 Daniel Stenberg (11 Dec 2008)
2130 - Internet Explorer had a broken HTTP digest authentication before v7 and
2131 there are servers "out there" that relies on the client doing this broken
2132 Digest authentication. Apache even comes with an option to work with such
2135 The difference is only for URLs that contain a query-part (a '?'-letter and
2136 text to the right of it).
2138 libcurl now supports this quirk, and you enable it by setting the
2139 CURLAUTH_DIGEST_IE bit in the bitmask you pass to the CURLOPT_HTTPAUTH or
2140 CURLOPT_PROXYAUTH options. They are thus individually controlled to server
2143 (note that there's no way to activate this with the curl tool yet)
2145 Daniel Fandrich (9 Dec 2008)
2146 - Added test cases 1089 and 1090 to test --write-out after a redirect to
2147 test a report that the size didn't work, but these test cases pass.
2149 - Documented CURLOPT_CONNECT_ONLY as being useful only on HTTP URLs.
2151 Daniel Stenberg (9 Dec 2008)
2152 - Ken Hirsch simplified how libcurl does FTPS: now it doesn't assume any
2153 particular state for the control connection like it did before for implicit
2154 FTPS (libcurl assumed such control connections to be encrypted while some
2155 FTPS servers such as FileZilla assumes such connections to be clear
2156 mode). Use the CURLOPT_USE_SSL option to set your desired level.
2158 Daniel Stenberg (8 Dec 2008)
2159 - Fred Machado posted about a weird FTP problem on the curl-users list and when
2160 researching it, it turned out he got a 550 response back from a SIZE command
2161 and then I fell over the text in RFC3659 that says:
2163 The presence of the 550 error response to a SIZE command MUST NOT be taken
2164 by the client as an indication that the file cannot be transferred in the
2165 current MODE and TYPE.
2167 In other words: the change I did on September 30th 2008 and that has been
2168 included in the last two releases were a regression and a bad idea. We MUST
2169 NOT take a 550 response from SIZE as a hint that the file doesn't exist.
2171 - Christian Krause filed bug #2221237
2172 (http://curl.haxx.se/bug/view.cgi?id=2221237) that identified an infinite
2173 loop during GSS authentication given some specific conditions. With his
2174 patience and great feedback I managed to narrow down the problem and
2175 eventually fix it although I can't test any of this myself!
2177 Daniel Fandrich (3 Dec 2008)
2178 - Fixed the getifaddrs version of Curl_if2ip to work on systems without IPv6
2179 support (e.g. Minix)
2181 Daniel Stenberg (3 Dec 2008)
2182 - Igor Novoseltsev filed bug #2351645
2183 (http://curl.haxx.se/bug/view.cgi?id=2351645) that identified a problem with
2184 the multi interface that occured if you removed an easy handle while in
2185 progress and the handle was used in a HTTP pipeline.
2187 - Pawel Kierski pointed out a mistake in the cookie code that could lead to a
2188 bad fclose() after a fatal error had occured.
2189 (http://curl.haxx.se/bug/view.cgi?id=2382219)
2191 Daniel Fandrich (25 Nov 2008)
2192 - If a HTTP request is Basic and num is already >=1000, the HTTP test
2193 server adds 1 to num to get the data section to return. This allows
2194 testing authentication negotiations using the Basic authentication
2197 - Added tests 1087 and 1088 to test Basic authentication on a redirect
2198 with and without --location-trusted
2200 Daniel Stenberg (24 Nov 2008)
2201 - Based on a patch by Vlad Grachov, libcurl now uses a new libssh2 0.19
2202 function when built to support SCP and SFTP that helps the library to know
2203 in which direction a particular libssh2 operation would return EAGAIN so
2204 that libcurl knows what socket conditions to wait for before trying the
2205 function call again. Previously (and still when using libssh2 0.18 or
2206 earlier), libcurl will busy-loop in this situation when the easy interface
2209 Daniel Fandrich (20 Nov 2008)
2210 - Automatically detect OpenBSD's CA cert bundle.
2212 Daniel Stenberg (19 Nov 2008)
2213 - I removed the default use of "Pragma: no-cache" from libcurl when a proxy is
2214 used. It has been used since forever but it was never a good idea to use
2215 unless explicitly asked for.
2217 - Josef Wolf's extension that allows a $TESTDIR/gdbinit$testnum file that when
2218 you use runtests.pl -g, will be sourced by gdb to allow additional fancy or
2219 whatever you see fit
2221 - Christian Krause reported and fixed a memory leak that would occur with HTTP
2222 GSS/kerberos authentication (http://curl.haxx.se/bug/view.cgi?id=2284386)
2224 - Andreas Wurf and Markus Koetter helped me analyze a problem that Andreas got
2225 when uploading files to a single FTP server using multiple easy handle
2226 handles with the multi interface. Occasionally a handle would stall in
2229 The problem turned out to be a side-effect of the ConnectionExists()
2230 function's eagerness to re-use a handle for HTTP pipelining so it would
2231 select it even if already being in use, due to an inadequate check for its
2232 chances of being used for pipelnining.
2234 Daniel Fandrich (17 Nov 2008)
2235 - Added more compiler warning options for gcc 4.3
2237 Yang Tse (17 Nov 2008)
2238 - Fix a remaining problem in the inet_pton() runtime configure check. And
2239 fix internal Curl_inet_pton() failures to reject certain malformed literals.
2241 - Make configure script check if ioctl with the SIOCGIFADDR command can be
2242 used, and define HAVE_IOCTL_SIOCGIFADDR if appropriate.
2244 Daniel Stenberg (16 Nov 2008)
2245 - Christian Krause fixed a build failure when building with gss support
2246 enabled and FTP disabled.
2248 - Added check for NULL returns from strdup() in src/main.c and lib/formdata.c
2249 - reported by Jim Meyering also prevent buffer overflow on MSDOS when you do
2250 for example -O on a url with a file name part longer than PATH_MAX letters
2252 - lib/nss.c fixes based on the report by Jim Meyering: I went over and added
2253 checks for return codes for all calls to malloc and strdup that were
2254 missing. I also changed a few malloc(13) to use arrays on the stack and a
2255 few malloc(PATH_MAX) to instead use aprintf() to lower memory use.
2257 - I fixed a memory leak in Curl_nss_connect() when CURLOPT_ISSUERCERT is
2260 Daniel Fandrich (14 Nov 2008)
2261 - Added .xml as one of the few common file extensions known by the multipart
2264 - Added some #ifdefs around header files and change the EAGAIN test to
2265 fix compilation on Cell (reported by Jeff Curley).
2267 Yang Tse (14 Nov 2008)
2268 - Fixed several configure script issues affecting checks for inet_ntoa_r(),
2269 inet_ntop(), inet_pton(), getifaddrs(), fcntl() and getaddrinfo().
2271 Yang Tse (13 Nov 2008)
2272 - Refactored configure script detection of functions used to set sockets into
2273 non-blocking mode, and decouple function detection from function capability.