5 \___|\___/|_| \_\_____|
9 Daniel Stenberg (5 June 2010)
10 - Constantine Sapuntzakis fixed a case of spurious SSL connection aborts using
11 libcurl and OpenSSL. "I tracked it down to uncleared error state on the
12 OpenSSL error stack - patch attached deals with that."
14 Daniel Stenberg (5 June 2010)
15 - Frank Meier added CURLINFO_PRIMARY_PORT, CURLINFO_LOCAL_IP and
16 CURLINFO_LOCAL_PORT to curl_easy_getinfo().
18 Yang Tse (4 June 2010)
19 - Enabled OpenLDAP support for cygwin builds. This support was disabled back
20 in 2008 due to incompatibilities between OpenSSL and OpenLDAP headers.
21 cygwin's OpenSSL 0.9.8l and OpenLDAP 2.3.43 versions on cygwin 1.5.25
22 allow building an OpenLDAP enabled libcurl supporting back to Windows 95.
24 Removed the non-functional CURL_LDAP_HYBRID code and references.
26 Daniel Stenberg (2 June 2010)
27 - Jason McDonald posted bug report #3006786 when he found that the SFTP code
28 didn't timeout properly in several places in the code even if a timeout was
31 Based on his suggested patch, I wrote a different implementation that I
32 think addressed the issue better and also uses the connect timeout for the
33 initial part of the SSH/SFTP done during the "protocol connect" phase.
35 (http://curl.haxx.se/bug/view.cgi?id=3006786)
37 Yang Tse (2 June 2010)
38 - Added missing new libcurl files to non-configure targets. Adjusted
39 libcurl standard internal header inclusions in new files. Fixed an
40 SPNEGO related memory leak. Fixed several LDAP related compilation
41 issues, and fixed some compiler warnings.
43 Daniel Stenberg (1 June 2010)
44 - Igor Novoseltsev reported a problem with the multi socket API and using
45 timeouts and timers. It boiled down to a problem with libcurl's use of
46 GetTickCount() interally to figure out the current time, while Igor's own
47 application code used another function call.
49 It made his app call the socket API timeout function a bit _before_ libcurl
50 would consider the timeout to trigger, and that could easily lead to
51 timeouts or stalls in the app. It seems GetTickCount() in general often has
52 no better resolution than 16ms and switching to the alternative function
53 QueryPerformanceCounter has its share of problems:
54 http://www.virtualdub.org/blog/pivot/entry.php?id=106
56 We address this problem by simply having libcurl treat timers that already
57 has occured or will occur within 40ms subject for treatment. I'm confident
58 that there are other implementations and operating systems with similarly in
59 accurate timer functions so it makes sense to have applied generically and I
60 don't believe we sacrifice much by adding a 40ms inaccuracy on these
63 Kamil Dudka (27 May 2010)
64 - added a new test for CRL support (test313)
66 - Tor Arntsen changed the alternative definition of bool to use enum instead
69 Daniel Stenberg (25 May 2010)
70 - Julien Chaffraix fixed the warning seen when compiling lib/rtmp.c: one
71 unused variables, several unused arguments and some missing #include.
73 - Julien Chaffraix fixed 2 OOM errors: a missing NULL-check in
74 lib/http_negociate.c and a potential NULL dereferencing in lib/splay.c
76 - Howard Chu brought a patch that makes the LDAP code much cleaner, nicer and
77 in general being a better libcurl citizen. If a new enough OpenLDAP version
78 is detect, the new and shiny lib/openldap.c code is then used instead of the
81 Daniel Stenberg (21 May 2010)
82 - Eric Mertens posted bug #3003705: when we made TFTP use the correct timeout
83 option when sent to the server (fixed May 18th 2010) it became obvious that
84 libcurl used invalid timeout values (300 by default while the RFC allows
85 nothing above 255). While of course it is obvious that as TFTP has worked
86 thus far without being able to set timeout at all, just removing the setting
87 wouldn't make any difference in behavior. I decided to still keep it (but
88 fix the problem) as it now actually allows for easier (future) customization
91 (http://curl.haxx.se/bug/view.cgi?id=3003705)
93 - Douglas Kilpatrick filed bug report #3004787 and pointed out that the TFTP
94 code didn't handle block id wraps correctly. His suggested fix inspired the
97 (http://curl.haxx.se/bug/view.cgi?id=3004787)
99 Daniel Stenberg (20 May 2010)
100 - Tanguy Fautre brought a fix to allow curl to build with Microsoft VC10.
102 Daniel Stenberg (18 May 2010)
103 - Eric Mertens posted bug report #3003005 pointing out that the libcurl TFTP
104 code was not sending the timeout option properly to the server, and
107 (http://curl.haxx.se/bug/view.cgi?id=3003005)
109 Kamil Dudka (16 May 2010)
110 - Pavel Raiskup introduced a new option CURLOPT_FNMATCH_DATA in order to pass
111 a custom data pointer to the callback specified by CURLOPT_FNMATCH_FUNCTION.
113 Daniel Stenberg (14 May 2010)
114 - John-Mark Bell filed bug #3000052 that identified a problem (with an
115 associated patch) with the OpenSSL handshake state machine when the multi
118 Performing an https request using a curl multi handle and using select or
119 epoll to wait for events results in a hang. It appears that the cause is the
120 fix for bug #2958179, which makes ossl_connect_common unconditionally return
121 from the step 2 loop when fetching from a multi handle.
123 When ossl_connect_step2 has completed, it updates connssl->connecting_state
124 to ssl_connect_3. ossl_connect_common will then return to the caller, as a
125 multi handle is in use. Eventually, the client code will call
126 curl_multi_fdset to obtain an updated fdset to select or epoll on. For https
127 requests, curl_multi_fdset will cause https_getsock to be called.
128 https_getsock will only return a socket handle if the connecting_state is
129 ssl_connect_2_reading or ssl_connect_2_writing. Therefore, the client will
130 never obtain a valid fdset, and thus not drive the multi handle, resulting
133 (http://curl.haxx.se/bug/view.cgi?id=3000052)
135 - Sebastian V reported bug #3000056 identifying a problem with redirect
136 following. It showed that when curl followed redirects it didn't properly
137 ignore the response body of the 30X response if that response was using
138 compressed Content-Encoding!
140 (http://curl.haxx.se/bug/view.cgi?id=3000056)
142 Daniel Stenberg (12 May 2010)
143 - Howard Chu brought support for RTMP. This is powered by the underlying
144 librtmp library. It supports a range of variations and "sub-protocols"
145 within the RTMP family.
147 - Pavel Raiskup brought support for FTP directory wildcard matching to allow
148 selective downloading. To provide that, a set of new options were added:
150 CURLOPT_WILDCARDMATCH
151 CURLOPT_CHUNK_BGN_FUNCTION
152 CURLOPT_CHUNK_END_FUNCTION
154 CURLOPT_FNMATCH_FUNCTION
156 There were also a set of new tests added (574 - 577) to verify this.
158 Kamil Dudka (11 May 2010)
159 - CRL support in libcurl-NSS has been completely broken. Now it works. Original
160 bug report: https://bugzilla.redhat.com/581926
162 Daniel Stenberg (7 May 2010)
163 - Dirk Manske reported a regression. When connecting with the multi interface,
164 there were situations where libcurl wouldn't store connect time correctly as
165 it used to (and is documented to) do.
167 Using his fine sample program we could repeat it, and I wrote up test case
168 573 using that code. The problem does not easily show itself using the local
171 The fix, also as suggested by Dirk, is a bit on the ugly side as it adds yet
172 another call to Curl_verboseconnect() and setting the TIMER_CONNECT time.
173 That situation is subject for some closer inspection in the future.
175 - Howard Chu split the I/O handling functions into private handlers.
177 Howard Chu brought the bulk work of this patch that properly moves out the
178 sending and recving of data to the parts of the code that are properly
179 responsible for the various ways of doing so.
181 Daniel Stenberg assisted with polishing a few bits and fixed some minor
182 flaws in the original patch.
184 Another upside of this patch is that we now abuse CURLcodes less with the
185 "magic" -1 return codes and instead use CURLE_AGAIN more consistently.
187 Daniel Stenberg (5 May 2010)
188 - Hoi-Ho Chan introduced support for using the PolarSSL library. You control
189 this with the new configure option --with-polarssl.
191 Daniel Stenberg (29 Apr 2010)
192 - Ben Greear made telnet a lot better/easier to use by an application:
194 The main change is to allow input from user-specified methods, when they are
195 specified with CURLOPT_READFUNCTION. All calls to fflush(stdout) in
196 telnet.c were removed, which makes using 'curl telnet://foo.com' painful
197 since prompts and other data are not always returned to the user promptly.
198 Use 'curl --no-buffer telnet://foo.com' instead. In general, the user
199 should have their CURLOPT_WRITEFUNCTION do a fflush for interactive use.
201 Also fix assumption that reading from stdin never returns < 0.
202 Old code could crash in that case.
204 Call progress functions in telnet main loop.
206 Daniel Stenberg (26 Apr 2010)
207 - Make use of the libssh2_init/exit functions that libssh2 added in version
208 1.2.5. Using them will improve how libcurl works in threaded situations when
209 SCP and SFTP are transfered.
211 Daniel Stenberg (25 Apr 2010)
212 - Based on work by Kamil Dudka, I've introduced the new configure option
213 --enable-threaded-resolver. When used, the configure script will check for
214 pthreads and if around, it will build libcurl to use pthreads to do name
215 resolving in a threaded manner. Note that this is just a fix to offer an
216 option that can enable the code that already included. The threader resolver
217 code was mostly added on Jan 26 2010.
219 Daniel Stenberg (24 Apr 2010)
220 - Alex Bligh introduced the --proto and -proto-redir options that limit what
221 protocols curl accepts for the requests and when following redirects.
223 Kamil Dudka (24 Apr 2010)
224 - Fixed test536 in order to not fail with threaded DNS resolver and tweaked
225 comments in certain examples using curl_multi_fdset().
227 - Fixed SSL handshake timeout underflow in libcurl-NSS, which caused test405
228 to hang on a slow machine.
230 Daniel Stenberg (21 Apr 2010)
231 - The -O option caused curl to crash on windows and DOS due to the tool
232 writing out of boundary memory.
234 Yang Tse (20 Apr 2010)
235 - Ruslan Gazizov detected that MSVC makefiles were using wsock32.lib instead
236 of ws2_32.lib, this generated linking issues on MSVC IPv6 enabled builds
237 that were done using those makefiles.
239 Daniel Stenberg (19 Apr 2010)
240 - -J/--remote-header-name didn't strip trailing carriage returns or linefeeds
241 properly, so they could be used in the file name.
243 Daniel Stenberg (16 Apr 2010)
244 - Jerome Vouillon made the GnuTLS SSL handshake phase non-blocking.
246 - The recent overhaul of the SSL recv function made the GnuTLS specific code
247 treat a zero returned from gnutls_record_recv() as an error, and this caused
248 our HTTPS test cases to fail. We leave it to upper layer code to detect if
249 an EOF is a problem or not.
251 - I reverted the resolver fix from yesterday and instead removed all uses of
252 AI_CANONNAME all over libcurl and made the only user of that info (krb5.c)
253 use the host name from the URL instead. No reverse resolving is a good
256 - Paul Howarth made configure properly detect GSS "on ancient Linux distros"
257 by editing in which order we use headers to detect GSS.
259 Daniel Stenberg (15 Apr 2010)
260 - Rainer Canavan filed bug report #2987196 that identified libcurl doing
261 unnecesary reverse name lookups in many cases when built to use IPv4 and
262 getaddrinfo(). The logic for ipv6 is now used for ipv4 too.
264 (http://curl.haxx.se/bug/view.cgi?id=2963679)
266 Version 7.20.1 (14 April 2010)
268 Daniel Stenberg (9 Apr 2010)
269 - Prefixing the FTP quote commands with an asterisk really only worked for the
270 postquote actions. This is now fixed and test case 227 has been extended to
273 Kamil Dudka (4 Apr 2010)
274 - Eliminated a race condition in Curl_resolv_timeout().
276 - Refactorized interface of Curl_ssl_recv()/Curl_ssl_send().
278 - libcurl-NSS now provides more accurate messages and error codes in case of
279 client certificate problem. Either during connection, or transfer phase.
281 Daniel Stenberg (1 Apr 2010)
282 - Matt Wixson found and fixed a bug in the SCP/SFTP area where the code
283 treated a 0 return code from libssh2 to be the same as EAGAIN while in
284 reality it isn't. The problem caused a hang in SFTP transfers from a
287 Daniel Stenberg (28 Mar 2010)
288 - Ben Greear: If you pass a URL to pop3 that does not contain a message ID as
289 part of the URL, it would previously ask for 'INBOX' which just causes the
290 pop3 server to return an error.
292 Now libcurl treats en empty message ID as a request for LIST (list of pop3
293 message IDs). User's code could then parse this and download individual
296 Daniel Stenberg (27 Mar 2010)
297 - Ben Greear brought a patch that from now on allows all protocols to specify
298 name and user within the URL, in the same manner HTTP and FTP have been
299 allowed to in the past - although far from all of the libcurl supported
300 protocls actually have that feature in their URL definition spec.
302 Daniel Stenberg (26 Mar 2010)
303 - Ben Greear brought code that makes the rate limiting code for the easy
304 interface a bit smoother as it introduces sub-second sleeps during it and it
305 also takes the buffer sizes into account.
307 Daniel Stenberg (24 Mar 2010)
308 - Bob Richmond: There's an annoying situation where libcurl will read new HTTP
309 response data from a socket, then check if it's a timeout if one is set. If
310 the last packet received constitutes the end of the response body, libcurl
311 still treats it as a timeout condition and reports a message like:
313 "Operation timed out after 3000 milliseconds with 876 out of 876 bytes
316 It should only a timeout if the timer lapsed and we DIDN'T receive the end
317 of the response body yet.
319 - Christopher Conroy fixed a problem with RTSP and GET_PARAMETER reported
320 to us by Massimo Callegari. There's a new test case 572 that verifies this
323 - The 'ares' subtree has been removed from the source repository. It was
324 always a separate project that sort of piggybacked on the curl project since
325 the dawn of times and now the time has come for it to go stand on its own
326 legs and continue living its own life. All details on c-ares and its new
327 source code repository is found at http://c-ares.haxx.se/
329 Daniel Stenberg (23 Mar 2010)
330 - Kenny To filed the bug report #2963679 with patch to fix a problem he
331 experienced with doing multi interface HTTP POST over a proxy using
332 PROXYTUNNEL. He found a case where it would connect fine but bits.tcpconnect
333 was not set correct so libcurl didn't work properly.
335 (http://curl.haxx.se/bug/view.cgi?id=2963679)
337 - Akos Pasztory filed debian bug report #572276
338 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=572276 mentioning a problem
339 with a resource that returns chunked-encoded _and_ with a Content-Length
340 and libcurl failed to properly ignore the latter information.
342 - Hauke Duden provided an example program that made the multi interface crash.
343 His example simply used the multi interface and did first one FTP transfer
344 and after completion it used a second easy handle and did another FTP
345 transfer on the same FTP server.
347 This triggered a bug in the "delayed easy handle kill" system that curl
348 uses: when an FTP connection is left alive it must keep an easy handle
349 around internally - only for the purpose of having an easy handle when it
350 later disconnects it. The code assumed that when the easy handle was removed
351 and an internal reference was made, that version could be killed later on
352 when a new easy handle came using the same connection. This was wrong as
353 Hauke's example showed that the removed handle wasn't killed for real until
354 later. This caused a double close attempt => segfault.
356 Daniel Stenberg (22 Mar 2010)
357 - Thomas Lopatic fixed the alarm()-based DNS timeout:
359 Looking at the code of Curl_resolv_timeout() in hostip.c, I think that in
360 case of a timeout, the signal handler for SIGALRM never gets removed. I
361 think that in my case it gets executed at some point later on when execution
362 has long left Curl_resolv_timeout() or even the cURL library.
364 The code that is jumped to with siglongjmp() simply sets the error message
365 to "name lookup timed out" and then returns with CURLRESOLV_ERROR. I guess
366 that instead of simply returning without cleaning up, the code should have a
367 goto that jumps to the spot right after the call to Curl_resolv().
369 Kamil Dudka (22 Mar 2010)
370 - Douglas Steinwand contributed a patch fixing insufficient initialization in
371 Curl_clone_ssl_config()
373 Daniel Stenberg (21 Mar 2010)
374 - Ben Greear improved TFTP: the error code returning and the treatment
375 of TSIZE == 0 when uploading.
377 - We've switched from CVS to git. See http://curl.haxx.se/source.html
379 Kamil Dudka (19 Mar 2010)
380 - Improved Curl_read() to not ignore the error returned from Curl_ssl_recv().
382 Daniel Stenberg (15 Mar 2010)
383 - Constantine Sapuntzakis brought a patch:
385 The problem mentioned on Dec 10 2009
386 (http://curl.haxx.se/bug/view.cgi?id=2905220) was only partially fixed.
387 Partially because an easy handle can be associated with many connections in
388 the cache (e.g. if there is a redirect during the lifetime of the easy
389 handle). The previous patch only cleaned up the first one. The new fix now
390 removes the easy handle from all connections, not just the first one.
392 Daniel Stenberg (6 Mar 2010)
393 - Ben Greear brought a patch that fixed the rate limiting logic for TFTP when
394 the easy interface was used.
396 Daniel Stenberg (5 Mar 2010)
397 - Daniel Johnson provided fixes for building curl with the clang compiler.
399 Yang Tse (5 Mar 2010)
400 - Constantine Sapuntzakis detected and fixed a double free in builds done
401 with threaded resolver enabled (Windows default configuration) that would
402 get triggered when a curl handle is closed while doing DNS resolution.
404 Daniel Stenberg (2 Mar 2010)
405 - [Daniel Johnson] I've been trying to build libcurl with clang on Darwin and
406 ran into some issues with the GSSAPI tests in configure.ac. The tests first
407 try to determine the include dirs and libs and set CPPFLAGS and LIBS
408 accordingly. It then checks for the headers and finally sets LIBS a second
409 time, causing the libs to be included twice. The first setting of LIBS seems
410 redundant and should be left out, since the first part is otherwise just
411 about finding headers.
413 My second issue is that 'krb5-config --libs gssapi' on Darwin is less than
414 useless and returns junk that, while it happens to work with gcc, causes
415 clang to choke. For example, --libs returns $CFLAGS along with the libs,
416 which is really retarded. Simply setting 'LIBS="$LIBS -lgssapi_krb5
417 -lresolv"' on Darwin is sufficient.
419 - Based on patch provided by Jacob Moshenko, the transfer logic now properly
420 makes sure that when using sub-second timeouts, there's no final bad 1000ms
421 wait. Previously, a sub-second timeout would often make the elapsed time end
422 up the time rounded up to the nearest second (e.g. 1s for 200ms timeout)
424 - Andrei Benea filed bug report #2956698 and pointed out that the
425 CURLOPT_CERTINFO feature leaked memory due to a missing OpenSSL function
426 call. He provided the patch to fix it too.
428 http://curl.haxx.se/bug/view.cgi?id=2956698
430 - Markus Duft pointed out in bug #2961796 that even though Interix has a
431 poll() function it doesn't quite work the way we want it so we must disable
432 it, and he also provided a patch for it.
434 http://curl.haxx.se/bug/view.cgi?id=2961796
436 - Made the pingpong timeout code properly deal with the response timeout AND
437 the global timeout if set. Also, as was reported in the bug report #2956437
438 by Ryan Chan, the time stamp to use as basis for the per command timeout was
439 not set properly in the DONE phase for FTP (and not for SMTP) so I fixed
440 that just now. This was a regression compared to 7.19.7 due to the
441 conversion of FTP code over to the generic pingpong concepts.
443 http://curl.haxx.se/bug/view.cgi?id=2956437
445 Daniel Stenberg (1 Mar 2010)
446 - Ben Greear provided an update for TFTP that fixes upload.
448 - Wesley Miaw reported bug #2958179 which identified a case of looping during
449 OpenSSL based SSL handshaking even though the multi interface was used and
450 there was no good reason for it.
452 http://curl.haxx.se/bug/view.cgi?id=2958179
454 Daniel Stenberg (26 Feb 2010)
455 - Pat Ray in bug #2958474 pointed out an off-by-one case when receiving a
456 chunked-encoding trailer.
458 http://curl.haxx.se/bug/view.cgi?id=2958474
460 Daniel Fandrich (25 Feb 2010)
461 - Fixed a couple of out of memory leaks and a segfault in the SMTP & IMAP code.
463 Yang Tse (25 Feb 2010)
464 - I fixed bug report #2958074 indicating
465 (http://curl.haxx.se/bug/view.cgi?id=2958074) that curl on Windows with
466 option --trace-time did not use local time when timestamping trace lines.
467 This could also happen on other systems depending on time souurce.
469 Patrick Monnerat (22 Feb 2010)
470 - Proper handling of STARTTLS on SMTP, taking CURLUSESSL_TRY into account.
471 - SMTP falls back to RFC821 HELO when EHLO fails (and SSL is not required).
472 - Use of true local host name (i.e.: via gethostname()) when available, as
473 default argument to SMTP HELO/EHLO.
474 - Test case 804 for HELO fallback.
476 Daniel Stenberg (20 Feb 2010)
477 - Fixed the SMTP compliance by making sure RCPT TO addresses are specified
478 properly in angle brackets. Recipients provided with CURLOPT_MAIL_RCPT now
479 get angle bracket wrapping automatically by libcurl unless the recipient
480 starts with an angle bracket as then the app is assumed to deal with that
483 - I made the SMTP code expect a 250 response back from the server after the
484 full DATA has been sent, and I modified the test SMTP server to also send
485 that response. As usual, the DONE operation that is made after a completed
486 transfer is still not doable in a non-blocking way so this waiting for 250
487 is unfortunately made blockingly.
489 Yang Tse (14 Feb 2010)
490 - Overhauled test suite getpart() function. Fixing potential out of bounds
491 stack and memory overwrites triggered with huge test case definitions.
493 Daniel Stenberg (13 Feb 2010)
494 - Martin Hager reported and fixed a problem with a missing quote in libcurl.m4
496 (http://curl.haxx.se/bug/view.cgi?id=2951319)
498 - Tom Donovan fixed the CURL_FORMAT_* defines when building with cmake.
500 (http://curl.haxx.se/bug/view.cgi?id=2951269)
502 Daniel Stenberg (12 Feb 2010)
503 - Jack Zhang reported a problem with SMTP: we wrongly used multiple addresses
504 in the same RCPT TO line, when they should be sent in separate single
505 commands. I updated test case 802 to verify this.
507 - I also fixed a bad use of my_setopt_str() of CURLOPT_MAIL_RCPT in the curl
508 tool which made it try to output it as string for the --libcurl feature
509 which could lead to crashes.
511 Yang Tse (11 Feb 2010)
512 - Steven M. Schweda fixed VMS builder bad behavior when used in a batch job,
513 removed obsolete batch_compile.com and defines.com and updated VMS readme.
515 Version 7.20.0 (9 February 2010)
517 Daniel Stenberg (9 Feb 2010)
518 - When downloading compressed content over HTTP and the app asked libcurl to
519 automatically uncompress it with the CURLOPT_ENCODING option, libcurl could
520 wrongly provide the callback with more data than the maximum documented
521 amount. An application could thus get tricked into badness if the maximum
522 limit was trusted to be enforced by libcurl itself (as it is documented).
524 This is further detailed and explained in the libcurl security advisory
527 http://curl.haxx.se/docs/adv_20100209.html
529 Daniel Fandrich (3 Feb 2010)
530 - Changed the Watcom makefiles to make them easier to keep in sync with
531 Makefile.inc since that can't be included directly.
533 Yang Tse (2 Feb 2010)
534 - Symbol CURL_FORMAT_OFF_T now obsoleted, will be removed in a future release,
535 symbol will not be available when building with CURL_NO_OLDIES defined. Use
536 of CURL_FORMAT_CURL_OFF_T is preferred since 7.19.0
538 Daniel Stenberg (1 Feb 2010)
539 - Using the multi_socket API, it turns out at times it seemed to "forget"
540 connections (which caused a hang). It turned out to be an existing (7.19.7)
541 bug in libcurl (that's been around for a long time) and it happened like
544 The app calls curl_multi_add_handle() to add a new easy handle, libcurl will
545 then set it to timeout in 1 millisecond so libcurl will tell the app about
548 The app's timeout fires off that there's a timeout, the app calls libcurl as
549 we so often document it:
552 res = curl_multi_socket_action(... TIMEOUT ...);
553 } while(CURLM_CALL_MULTI_PERFORM == res);
555 And this is the problem number one:
557 When curl_multi_socket_action() is called with no specific handle, but only
558 a timeout-action, it will *only* perform actions within libcurl that are
559 marked to run at this time. In this case, the request would go from INIT to
560 CONNECT and return CURLM_CALL_MULTI_PERFORM. When the app then calls libcurl
561 again, there's no timer set for this handle so it remains in the CONNECT
562 state. The CONNECT state is a transitional state in libcurl so it reports no
563 sockets there, and thus libcurl never tells the app anything more about that
564 easy handle/connection.
566 libcurl _does_ set a 1ms timeout for the handle at the end of
567 multi_runsingle() if it returns CURLM_CALL_MULTI_PERFORM, but since the loop
568 is instant the new job is not ready to run at that point (and there's no
569 code that makes libcurl call the app to update the timout for this new
570 timeout). It will simply rely on that some other timeout will trigger later
571 on or that something else will update the timeout callback. This makes the
572 bug fairly hard to repeat.
574 The fix made to adress this issue:
576 We introduce a loop in lib/multi.c around all calls to multi_runsingle() and
577 simply check for CURLM_CALL_MULTI_PERFORM internally. This has the added
578 benefit that this goes in line with my long-term wishes to get rid of the
579 CURLM_CALL_MULTI_PERFORM all together from the public API.
581 The downside of this fix, is that the counter we return in 'running_handles'
582 in several of our public functions then gets a slightly new and possibly
583 confusing behavior during times:
585 If an app adds a handle that fails to connect (very quickly) it may just
586 as well never appear as a 'running_handle' with this fix. Previously it
587 would first bump the counter only to get it decreased again at next call.
588 Even I have used that change in handle counter to signal "end of a
589 transfer". The only *good* way to find the end of a individual transfer
590 is calling curl_multi_info_read() to see if it returns one.
592 Of course, if the app previously did the looping before it checked the
593 counter, it really shouldn't be any new effect.
595 Yang Tse (26 Jan 2010)
596 - Constantine Sapuntzakis' and Joshua Kwan's work done in the last four months
597 relative to the asynchronous DNS lookups, along with with some integration
598 adjustments I have done are finally committed to CVS.
600 Currently these enhancements will benefit builds done using c-ares on any
601 platform as well as Windows builds using the default threaded resolver.
603 This release does not make generally available POSIX threaded DNS lookups
604 yet. There is no configure option to enable this feature yet. It is possible
605 to experimantally try this feature running configure with compiler flags that
606 make simultaneous definition of preprocessor symbols USE_THREADS_POSIX and
607 HAVE_PTHREAD_H, as well as whatever reentrancy compiler flags and linker ones
608 are required to link and properly use pthread_* functions on each platform.
610 Daniel Stenberg (26 Jan 2010)
611 - Mike Crowe made libcurl return CURLE_COULDNT_RESOLVE_PROXY when it is the
612 proxy that cannot be resolved when using c-ares. This matches the behaviour
613 when not using c-ares.
615 Björn Stenberg (23 Jan 2010)
616 - Added a new flag: -J/--remote-header-name. This option tells the
617 -O/--remote-name option to use the server-specified Content-Disposition
618 filename instead of extracting a filename from the URL.
620 Daniel Stenberg (21 Jan 2010)
621 - Chris Conroy brought support for RTSP transfers, and with it comes 8(!) new
622 libcurl options for controlling what to get and how to receive posssibly
623 interleaved RTP data.
625 Daniel Stenberg (20 Jan 2010)
626 - As was pointed out on the http-state mailing list, the order of cookies in a
627 HTTP Cookie: header _needs_ to be sorted on the path length in the cases
628 where two cookies using the same name are set more than once using
629 (overlapping) paths. Realizing this, identically named cookies must be
630 sorted correctly. But detecting only identically named cookies and take care
631 of them individually is harder than just to blindly and unconditionally sort
632 all cookies based on their path lengths. All major browsers also already do
633 this, so this makes our behavior one step closer to them in the cookie area.
635 Test case 8 was the only one that broke due to this change and I updated it
638 Daniel Stenberg (19 Jan 2010)
639 - David McCreedy brought a fix and a new test case (129) to make libcurl work
640 again when downloading files over FTP using ASCII and it turns out that the
641 final size of the file is not the same as the initial size the server
642 reported. This is very common since servers don't take the newline
643 conversions into account.
645 Kamil Dudka (14 Jan 2010)
646 - Suppressed side effect of OpenSSL configure checks, which prevented NSS from
647 being properly detected under certain circumstances. It had been caused by
648 strange behavior of pkg-config when handling PKG_CONFIG_LIBDIR. pkg-config
649 distinguishes among empty and non-existent environment variable in that case.
651 Daniel Stenberg (12 Jan 2010)
652 - Gil Weber reported a peculiar flaw with the multi interface when doing SFTP
653 transfers: curl_multi_fdset() would return -1 and not set and file
654 descriptors several times during a transfer of a single file. It turned out
655 to be due to two different flaws now fixed. Gil's excellent recipe helped me
658 Daniel Stenberg (11 Jan 2010)
659 - Made sure that the progress callback is repeatedly called at a regular
660 interval even during very slow connects.
662 - The tests/runtests.pl script now checks to see if the test case that runs is
663 present in the tests/data/Makefile.am and outputs a notice message on the
664 screen if not. Each test file has to be included in that Makefile.am to get
665 included in release archives and forgetting to add files there is a common
666 mistake. This is an attempt to make it harder to forget.
668 Daniel Stenberg (9 Jan 2010)
669 - Johan van Selst found and fixed a OpenSSL session ref count leak:
671 ossl_connect_step3() increments an SSL session handle reference counter on
672 each call. When sessions are re-used this reference counter may be
673 incremented many times, but it will be decremented only once when done (by
674 Curl_ossl_session_free()); and the internal OpenSSL data will not be freed
675 if this reference count remains positive. When a session is re-used the
676 reference counter should be corrected by explicitly calling
677 SSL_SESSION_free() after each consecutive SSL_get1_session() to avoid
678 introducing a memory leak.
680 (http://curl.haxx.se/bug/view.cgi?id=2926284)
682 Daniel Stenberg (7 Jan 2010)
683 - Make sure the progress callback is called repeatedly even during very slow
684 name resolves when c-ares is used for resolving.
686 Claes Jakobsson (6 Jan 2010)
687 - Julien Chaffraix fixed so that the fragment part in an URL is not sent
688 to the server anymore.
690 Kamil Dudka (3 Jan 2010)
691 - Julien Chaffraix eliminated a duplicated initialization in singlesocket().
693 Daniel Stenberg (2 Jan 2010)
694 - Make curl support --ssl and --ssl-reqd instead of the previous FTP-specific
695 versions --ftp-ssl and --ftp-ssl-reqd as these options are now used to
696 control SSL/TLS for IMAP, POP3 and SMTP as well in addition to FTP. The old
697 option names are still working but the new ones are the ones listed and
700 Daniel Stenberg (1 Jan 2010)
701 - Ingmar Runge enhanced libcurl's FTP engine to support the PRET command. This
702 command is a special "hack" used by the drftpd server, but even though it is
703 a custom extension I've deemed it fine to add to libcurl since this server
704 seems to survive and people keep using it and want libcurl to support
705 it. The new libcurl option is named CURLOPT_FTP_USE_PRET, and it is also
706 usable from the curl tool with --ftp-pret. Using this option on a server
707 that doesn't support this command will make libcurl fail.
709 I added test cases 1107 and 1108 to verify the functionality.
711 The PRET command is documented at
712 http://www.drftpd.org/index.php/Distributed_PASV
714 Yang Tse (30 Dec 2009)
715 - Steven M. Schweda improved VMS build system, and Craig A. Berry helped
716 with the patch and testing.
718 Daniel Stenberg (26 Dec 2009)
719 - Renato Botelho and Peter Pentchev brought a patch that makes the libcurl
720 headers work correctly even on FreeBSD systems before v8.
722 (http://curl.haxx.se/bug/view.cgi?id=2916915)
724 Daniel Stenberg (17 Dec 2009)
725 - David Byron fixed Curl_ossl_cleanup to actually call ENGINE_cleanup when
728 - Follow-up fix for the proxy fix I did for Jon Nelson's bug. It turned out I
729 was a bit too quick and broke test case 1101 with that change. The order of
730 some of the setups is sensitive. I now changed it slightly again to make
731 sure we do them in this order:
733 1 - parse URL and figure out what protocol is used in the URL
734 2 - prepend protocol:// to URL if missing
735 3 - parse name+password off URL, which needs to know what protocol is used
736 (since only some allows for name+password in the URL)
737 4 - figure out if a proxy should be used set by an option
738 5 - if no proxy option, check proxy environment variables
739 6 - run the protocol-specific setup function, which needs to have the proxy
742 Daniel Stenberg (15 Dec 2009)
743 - Jon Nelson found a regression that turned out to be a flaw in how libcurl
744 detects and uses proxies based on the environment variables. If the proxy
745 was given as an explicit option it worked, but due to the setup order
746 mistake proxies would not be used fine for a few protocols when picked up
747 from '[protocol]_proxy'. Obviously this broke after 7.19.4. I now also added
748 test case 1106 that verifies this functionality.
750 (http://curl.haxx.se/bug/view.cgi?id=2913886)
752 Daniel Stenberg (12 Dec 2009)
753 - IMAP, POP3 and SMTP support and their TLS versions (including IMAPS, POP3S
754 and SMTPS) are now supported. The current state may not yet be solid, but
755 the foundation is in place and the test suite has some initial support for
756 these protocols. Work will now persue to make them nice libcurl citizens
759 The work with supporting these new protocols was sponsored by
760 networking4all.com - thanks!
762 Daniel Stenberg (10 Dec 2009)
763 - Siegfried Gyuricsko found out that the curl manual said --retry would retry
764 on FTP errors in the transient 5xx range. Transient FTP errors are in the
765 4xx range. The code itself only tried on 5xx errors that occured _at login_.
766 Now the retry code retries on all FTP transfer failures that ended with a
769 (http://curl.haxx.se/bug/view.cgi?id=2911279)
771 - Constantine Sapuntzakis figured out a case which would lead to libcurl
772 accessing alredy freed memory and thus crash when using HTTPS (with
773 OpenSSL), multi interface and the CURLOPT_DEBUGFUNCTION and a certain order
774 of cleaning things up. I fixed it.
776 (http://curl.haxx.se/bug/view.cgi?id=2905220)
778 Daniel Stenberg (7 Dec 2009)
779 - Martin Storsjo made libcurl use the Expect: 100-continue header for posts
780 with unknown size. Previously it was only used for posts with a known size
781 larger than 1024 bytes.
783 Daniel Stenberg (1 Dec 2009)
784 - If the Expect: 100-continue header has been set by the application through
785 curl_easy_setopt with CURLOPT_HTTPHEADER, the library should set
786 data->state.expect100header accordingly - the current code (in 7.19.7 at
787 least) doesn't handle this properly. Martin Storsjo provided the fix!
789 Yang Tse (28 Nov 2009)
790 - Added Diffie-Hellman parameters to several test harness certificate files in
791 PEM format. Required by several stunnel versions used by our test harness.
793 Daniel Stenberg (28 Nov 2009)
794 - Markus Koetter provided a polished and updated version of Chad Monroe's TFTP
795 rework patch that now integrates TFTP properly into libcurl so that it can
796 be used non-blocking with the multi interface and more. BLKSIZE also works.
798 The --tftp-blksize option was added to allow setting the TFTP BLKSIZE from
801 Daniel Stenberg (26 Nov 2009)
802 - Extended and fixed the change I did on Dec 11 for the the progress
803 meter/callback during FTP command/response sequences. It turned out it was
804 really lame before and now the progress meter SHOULD get called at least
807 Daniel Stenberg (23 Nov 2009)
808 - Bjorn Augustsson reported a bug which made curl not report any problems even
809 though it failed to write a very small download to disk (done in a single
810 fwrite call). It turned out to be because fwrite() returned success, but
811 there was insufficient error-checking for the fclose() call which tricked
812 curl to believe things were fine.
814 Yang Tse (23 Nov 2009)
815 - David Byron modified Makefile.dist vc8 and vc9 targets in order to allow
816 finer granularity control when generating src and lib makefiles.
818 Yang Tse (22 Nov 2009)
819 - I modified configure to force removal of the curlbuild.h file included in
820 distribution tarballs for use by non-configure systems. As intended, this
821 would get overwriten when doing in-tree builds. But VPATH builds would end
822 having two curlbuild.h files, one in the source tree and another in the
823 build tree. With the modification I introduced 5 Nov 2009 this could become
824 an issue when running libcurl's test suite.
826 Daniel Stenberg (20 Nov 2009)
827 - Constantine Sapuntzakis identified a write after close, as the sockets were
828 closed by libcurl before the SSL lib were shutdown and they may write to its
829 socket. Detected to at least happen with OpenSSL builds.
831 - Jad Chamcham pointed out a bug with connection re-use. If a connection had
832 CURLOPT_HTTPPROXYTUNNEL enabled over a proxy, a subsequent request using the
833 same proxy with the tunnel option disabled would still wrongly re-use that
834 previous connection and the outcome would only be badness.
836 Yang Tse (18 Nov 2009)
837 - I modified the memory tracking system to make it intolerant with zero sized
838 malloc(), calloc() and realloc() function calls.
840 Daniel Stenberg (17 Nov 2009)
841 - Constantine Sapuntzakis provided another fix for the DNS cache that could
842 end up with entries that wouldn't time-out:
844 1. Set up a first web server that redirects (307) to a http://server:port
846 2. Have curl connect to the first web server using curl multi
848 After the curl_easy_cleanup call, there will be curl dns entries hanging
849 around with in_use != 0.
851 (http://curl.haxx.se/bug/view.cgi?id=2891591)
853 - Marc Kleine-Budde fixed: curl saved the LDFLAGS set during configure into
854 its pkg-config file. So -Wl stuff ended up in the .pc file, which is really
855 bad, and breaks if there are multiple -Wl in our LDFLAGS (which are in
856 PTXdist). bug #2893592 (http://curl.haxx.se/bug/view.cgi?id=2893592)
858 Kamil Dudka (15 Nov 2009)
859 - David Byron improved the configure script to use pkg-config to find OpenSSL
860 (and in particular the list of required libraries) even if a path is given
861 as argument to --with-ssl
863 Yang Tse (15 Nov 2009)
864 - I removed enable-thread / disable-thread configure option. These were only
865 placebo options. The library is always built as thread safe as possible on
868 Claes Jakobsson (14 Nov 2009)
869 - curl-config now accepts '--configure' to see what arguments was
870 passed to the configure script when building curl.
872 Daniel Stenberg (14 Nov 2009)
873 - Claes Jakobsson restored the configure functionality to detect NSS when
874 --with-nss is set but not "yes".
876 I think we can still improve that to check for pkg-config in that path etc,
877 but at least this patch brings back the same functionality we had before.
879 - Camille Moncelier added support for the file type SSL_FILETYPE_ENGINE for
880 the client certificate. It also disable the key name test as some engines
881 can select a private key/cert automatically (When there is only one key
882 and/or certificate on the hardware device used by the engine)
884 Yang Tse (14 Nov 2009)
885 - Constantine Sapuntzakis provided the fix that ensures that an SSL connection
886 won't be reused unless protection level for peer and host verification match.
888 I refactored how preprocessor symbol _THREAD_SAFE definition is done.
890 Kamil Dudka (12 Nov 2009)
891 - Kevin Baughman provided a fix preventing libcurl-NSS from crash on doubly
892 closed NSPR descriptor. The issue was hard to find, reported several times
893 before and always closed unresolved. More info at the RH bug:
894 https://bugzilla.redhat.com/534176
896 - libcurl-NSS now tries to reconnect with TLS disabled in case it detects
897 a broken TLS server. However it does not happen if SSL version is selected
898 manually. The approach was originally taken from PSM. Kaspar Brand helped me
899 to complete the patch. Original bug reports:
900 https://bugzilla.redhat.com/525496
901 https://bugzilla.redhat.com/527771
903 Yang Tse (12 Nov 2009)
904 - I modified configure script to make the getaddrinfo function check also
905 verify if the function is thread safe.
907 Yang Tse (11 Nov 2009)
908 - Marco Maggi reported that compilation failed when configured --with-gssapi
909 and GNU GSS installed due to a missing mutual exclusion of header files in
910 the Kerberos 5 code path. He also verified that my patch worked for him.
912 Daniel Stenberg (11 Nov 2009)
913 - Constantine Sapuntzakis posted bug #2891595
914 (http://curl.haxx.se/bug/view.cgi?id=2891595) which identified how an entry
915 in the DNS cache would linger too long if the request that added it was in
916 use that long. He also provided the patch that now makes libcurl capable of
917 still doing a request while the DNS hash entry may get timed out.
919 - Christian Schmitz noticed that the progress meter/callback was not properly
920 used during the FTP connection phase (after the actual TCP connect), while
921 it of course should be. I also made the speed check get called correctly so
922 that really slow servers will trigger that properly too.
924 Kamil Dudka (5 Nov 2009)
925 - Dropped misleading timeouts in libcurl-NSS and made sure the SSL socket works
926 in non-blocking mode.
928 Yang Tse (5 Nov 2009)
929 - I removed leading 'curl' path on the 'curlbuild.h' include statement in
930 curl.h, adjusting auto-makefiles include path, to enhance portability to
931 OS's without an orthogonal directory tree structure such as OS/400.
933 Daniel Stenberg (4 Nov 2009)
934 - I fixed several problems with the transfer progress meter. It showed the
935 wrong percentage for small files, most notable for <1000 bytes and could
936 easily end up showing more than 100% at the end. It also didn't show any
937 percentage, transfer size or estimated transfer times when transferring
940 Version 7.19.7 (4 November 2009)
942 Daniel Stenberg (2 Nov 2009)
943 - As reported independent by both Stan van de Burgt and Didier Brisebourg,
944 CURLINFO_SIZE_DOWNLOAD (the -w variable size_download) didn't work when
945 getting data from ldap!
947 Daniel Stenberg (31 Oct 2009)
948 - Gabriel Kuri reported a problem with CURLINFO_CONTENT_LENGTH_DOWNLOAD if the
949 download was 0 bytes, as libcurl would then return the size as unknown (-1)
950 and not 0. I wrote a fix and test case 566 to verify it.
952 Daniel Stenberg (30 Oct 2009)
953 - Liza Alenchery mentioned a problem with re-used SCP connection when a bad
954 auth is used, as it caused a crash. I failed to repeat the issue, but still
955 made a change that now forces the TCP connection used for a freed SCP
956 session to get closed and not be re-used.
958 - "Tom" posted a bug report that mentioned how libcurl did wrong when doing a
959 POST using a read callback, with Digest authentication and
960 "Transfer-Encoding: chunked" enforced. I would then cause the first request
961 to be wrongly sent and then basically hang until the server closed the
962 connection. I fixed the problem and added test case 565 to verify it.
964 Daniel Stenberg (25 Oct 2009)
965 - Dima Barsky made the curl cookie parser accept cookies even with blank or
966 unparsable expiry dates and then treat them as session cookies - previously
967 libcurl would reject cookies with a date format it couldn't parse. Research
968 shows that the major browser treat such cookies as session cookies. I
969 modified test 8 and 31 to verify this.
971 Daniel Stenberg (21 Oct 2009)
972 - Attempt to use pkg-config for finding out libssh2 installation details
975 - A patch in bug report #2883177 (http://curl.haxx.se/bug/view.cgi?id=2883177)
976 by Johan van Selst introduced the --crlfile option to curl, which makes curl
977 tell libcurl about a file with CRL (certificate revocation list) data to
980 Daniel Stenberg (18 Oct 2009)
981 - Ray Dassen provided a patch in Debian's bug tracker (bug number #551461)
982 that now makes curl_getdate(3) actually handles RFC 822 formatted dates that
983 use the "single letter military timezones".
984 http://www.rfc-ref.org/RFC-TEXTS/822/chapter5.html has the details.
986 - Fixed memory leak in the SCP/SFTP code as it never freed the knownhosts
989 - John Dennis filed bug report #2873666
990 (http://curl.haxx.se/bug/view.cgi?id=2873666) which identified a problem
991 which made libcurl loop infinitely when given incorrect credentials when
992 using HTTP GSS negotiate authentication. He also provided a small and simple
995 - Kevin Baughman found a double close() problem with libcurl-NSS, as when
996 libcurl called NSS to close the SSL "session" it also closed the actual
999 Yang Tse (17 Oct 2009)
1000 - Bug report #2866724 indicated
1001 (http://curl.haxx.se/bug/view.cgi?id=2866724) that curl on Windows failed
1002 when writing files whose file names originally contained characters which
1003 are not valid for file names on Windows. Dan Fandrich provided an initial
1004 patch and another revised one to fix this issue.
1006 Daniel Stenberg (1 Oct 2009)
1007 - Tom Mueller correctly reported in bug report #2870221
1008 (http://curl.haxx.se/bug/view.cgi?id=2870221) that libcurl returned an
1009 incorrect return code from the internal trynextip() function which caused
1010 him grief. This is a regression that was introduced in 7.19.1 and I find it
1011 strange it hasn't hit us harder, but I won't persue into figuring out
1014 - Constantine Sapuntzakis: The current implementation will always set
1015 SO_SNDBUF to CURL_WRITE_SIZE even if the SO_SNDBUF starts out larger. The
1016 patch doesn't do a setsockopt if SO_SNDBUF is already greater than
1017 CURL_WRITE_SIZE. This should help folks who have set up their computer with
1020 Daniel Stenberg (27 Sep 2009)
1021 - I introduced a maximum limit for received HTTP headers. It is controlled by
1022 the define CURL_MAX_HTTP_HEADER which is even exposed in the public header
1023 file to allow for users to fairly easy rebuild libcurl with a modified
1024 limit. The rationale for a fixed limit is that libcurl is realloc()ing a
1025 buffer to be able to put a full header into it, so that it can call the
1026 header callback with the entire header, but that also risk getting it into
1027 trouble if a server by mistake or willingly sends a header that is more or
1028 less without an end. The limit is set to 100K.
1030 Daniel Stenberg (26 Sep 2009)
1031 - John P. McCaskey posted a bug report that showed how libcurl did wrong when
1032 saving received cookies with no given path, if the path in the request had a
1033 query part. That is means a question mark (?) and characters on the right
1034 side of that. I wrote test case 1105 and fixed this problem.
1036 Kamil Dudka (26 Sep 2009)
1037 - Implemented a protocol independent way to specify blocking direction, used by
1038 transfer.c for blocking. It is currently used only by SCP and SFTP protocols.
1039 This enhancement resolves an issue with 100% CPU usage during SFTP upload,
1040 reported by Vourhey.
1042 Daniel Stenberg (25 Sep 2009)
1043 - Chris Mumford filed bug report #2861587
1044 (http://curl.haxx.se/bug/view.cgi?id=2861587) identifying that libcurl used
1045 the OpenSSL function X509_load_crl_file() wrongly and failed if it would
1046 load a CRL file with more than one certificate within. This is now fixed.
1048 Daniel Stenberg (16 Sep 2009)
1049 - Sven Anders reported that we introduced a cert verfication flaw for OpenSSL-
1050 powered libcurl in 7.19.6. If there was a X509v3 Subject Alternative Name
1051 field in the certficate it had to match and so even if non-DNS and non-IP
1052 entry was present it caused the verification to fail.
1054 Daniel Fandrich (15 Sep 2009)
1055 - Moved the libssh2 checks after the SSL library checks. This helps when
1056 statically linking since libssh2 needs the SSL library link flags to be
1057 set up already to satisfy its dependencies. This wouldn't be necessary if
1058 the libssh2 configure check was changed to use pkg-config since the
1059 --static flag would add the dependencies automatically.
1061 Yang Tse (14 Sep 2009)
1062 - Revert Joshua Kwan's patch committed 11 Sep 2009.
1064 Some systems poll function sets POLLHUP in revents without setting
1065 POLLIN, and sets POLLERR without setting POLLIN and POLLOUT. In some
1066 libcurl code execution paths this could trigger busy wait loops with
1067 high CPU usage until a timeout condition aborted the loop.
1069 The reverted patch addressed the above issue for a very specific case,
1070 when awaiting c-ares to resolve. A libcurl-wide fix for Curl_poll now
1071 superceeds this one.
1073 Guenter Knauf (11 Sep 2009)
1074 - Joshua Kwan provided a patch to pass POLLERR / POLLHUP back to c-ares.
1075 This fixes a loop problem with high CPU usage.
1077 Daniel Stenberg (10 Sep 2009)
1078 - Claes Jakobsson fixed a problem with cookie expiry dates at exctly the epoch
1079 start second "Thu Jan 1 00:00:00 GMT 1970" as the date parser then returns 0
1080 which internally then is treated as a session cookie. That particular date
1081 is now made to get the value of 1.
1083 Daniel Stenberg (2 Sep 2009)
1084 - Daniel Johnson found a flaw in the code converting sftp-errors to libcurl
1087 Daniel Stenberg (1 Sep 2009)
1088 - Peter Sylvester made a debug feature for Curl_resolv() that now will force
1089 libcurl to resolve 'localhost' whatever name you use in the URL *if* you set
1090 the --interface option to (exactly) "LocalHost". This will enable us to
1091 write tests for custom hosts names but still use a local host server.
1093 - configure now tries to use pkg-config for a number of sub-dependencies even
1094 when cross-compiling. The key to success is then you properly setup
1095 PKG_CONFIG_PATH before invoking configure.
1097 I also improved how NSS is detected by trying nss-config if pkg-config isn't
1098 present, and as a last resort just use the lib name and force the user to
1099 setup the LIBS/LDFLAGS/CFLAGS etc properly. The previous last resort would
1100 add a range of various libs that would almost never be quite correct.
1102 Daniel Stenberg (31 Aug 2009)
1103 - When using the multi interface with FTP and you asked for NOBODY, you did no
1104 QUOTE commands and the request used the same path as the connection had
1105 already changed to, it would decide that no commands would be necessary for
1106 the "DO" action and that was not handled properly but libcurl would instead
1109 Kamil Dudka (28 Aug 2009)
1110 - Improved error message for not matching certificate subject name in
1111 libcurl-NSS. Originally reported at:
1112 https://bugzilla.redhat.com/show_bug.cgi?id=516056#c9
1114 Patrick Monnerat (24 Aug 2009)
1115 - Introduced a SYST-based test to properly set-up name format when dealing
1116 with the OS/400 FTP server.
1118 - Fixed an ftp_readresp() bug preventing detection of failing control socket
1119 and causing FTP client to loop forever.
1121 Daniel Stenberg (24 Aug 2009)
1122 - Marc de Bruin pointed out that configure --with-gnutls=PATH didn't work
1123 properly and provided a fix. http://curl.haxx.se/bug/view.cgi?id=2843008
1125 - Eric Wong introduced support for the new option -T. (dot) that makes curl
1126 read stdin in a non-blocking fashion. This also brings back -T- (minus) to
1127 the previous blocking behavior since it could break stuff for people at
1130 Michal Marek (21 Aug 2009)
1131 - With CURLOPT_PROXY_TRANSFER_MODE, avoid sending invalid URLs like
1132 ftp://example.com;type=i if the user specified ftp://example.com without the
1135 Daniel Stenberg (21 Aug 2009)
1136 - Andre Guibert de Bruet pointed out a missing return code check for a
1137 strdup() that could lead to segfault if it returned NULL. I extended his
1138 suggest patch to now have Curl_retry_request() return a regular return code
1139 and better check that.
1141 - Lots of good work by Krister Johansen, mostly related to pipelining:
1143 Fix SIGSEGV on free'd easy_conn when pipe unexpectedly breaks
1144 Fix data corruption issue with re-connected transfers
1145 Fix use after free if we're completed but easy_conn not NULL
1147 Kamil Dudka (13 Aug 2009)
1148 - Changed NSS code to not ignore the value of ssl.verifyhost and produce more
1149 verbose error messages. Originally reported at:
1150 https://bugzilla.redhat.com/show_bug.cgi?id=516056
1152 Daniel Stenberg (12 Aug 2009)
1153 - Karl Moerder fixed the Makefile.vc* makefiles to include the new file
1154 nonblock.c so that they work fine again
1156 - I expanded test 517 with a bunch of more dates that originate from the
1157 Chrome browser test suite. It turns out most of them get parsed the same
1160 Version 7.19.6 (12 August 2009)
1162 Daniel Stenberg (12 Aug 2009)
1163 - Carsten Lange reported a bug and provided a patch for TFTP upload and the
1164 sending of the TSIZE option. I don't like fixing bugs just hours before
1165 a release, but since it was broken and the patch fixes this for him I decided
1166 to get it in anyway.
1168 Daniel Stenberg (11 Aug 2009)
1169 - Peter Sylvester made the HTTPS test server use specific certificates for
1170 each test, so that the test suite can now be used to actually test the
1171 verification of cert names etc. This made an error show up in the OpenSSL-
1172 specific code where it would attempt to match the CN field even if a
1173 subjectAltName exists that doesn't match. This is now fixed and verified
1176 - Benbuck Nason posted the bug report #2835196
1177 (http://curl.haxx.se/bug/view.cgi?id=2835196), fixing a few compiler
1178 warnings when mixing ints and bools.
1180 Daniel Fandrich (10 Aug 2009)
1181 - Fixed a memory leak in the FTP code and an off-by-one heap buffer overflow.
1183 Daniel Fandrich (9 Aug 2009)
1184 - Fixed some memory leaks in the command-line tool that caused most of the
1185 torture tests to fail.
1187 Daniel Stenberg (2 Aug 2009)
1188 - Curt Bogmine reported a problem with SNI enabled on a particular server. We
1189 should introduce an option to disable SNI, but as we're in feature freeze
1190 now I've addressed the obvious bug here (pointed out by Peter Sylvester): we
1191 shouldn't try to enable SNI when SSLv2 or SSLv3 is explicitly selected.
1192 Code for OpenSSL and GnuTLS was fixed. NSS doesn't seem to have a particular
1193 option for SNI, or are we simply not using it?
1195 Daniel Stenberg (1 Aug 2009)
1196 - Scott Cantor posted the bug report #2829955
1197 (http://curl.haxx.se/bug/view.cgi?id=2829955) mentioning the recent SSL cert
1198 verification flaw found and exploited by Moxie Marlinspike. The presentation
1199 he did at Black Hat is available here:
1200 https://www.blackhat.com/html/bh-usa-09/bh-usa-09-archives.html#Marlinspike
1202 Apparently at least one CA allowed a subjectAltName or CN that contain a
1203 zero byte, and thus clients that assumed they would never have zero bytes
1204 were exploited to OK a certificate that didn't actually match the site. Like
1205 if the name in the cert was "example.com\0theatualsite.com", libcurl would
1206 happily verify that cert for example.com.
1208 libcurl now better uses the length of the extracted name, not using the zero
1209 termination for getting the string length.
1211 This fixing only made and needed in OpenSSL interfacing code.
1213 - Tanguy Fautre pointed out that OpenSSL's function RAND_screen() (present
1214 only in some OpenSSL installs - like on Windows) isn't thread-safe and we
1215 agreed that moving it to the global_init() function is a decent way to deal
1216 with this situation.
1218 - Alexander Beedie provided the patch for a noproxy problem: If I have set
1219 CURLOPT_NOPROXY to "*", or to a host that should not use a proxy, I actually
1220 could still end up using a proxy if a proxy environment variable was set.
1222 Daniel Stenberg (27 Jul 2009)
1223 - All the quote options (CURLOPT_QUOTE, CURLOPT_POSTQUOTE and
1224 CURLOPT_PREQUOTE) now accept a preceeding asterisk before the command to
1225 send when using FTP, as a sign that libcurl shall simply ignore the response
1226 from the server instead of treating it as an error. Not treating a 400+ FTP
1227 response code as an error means that failed commands will not abort the
1228 chain of commands, nor will they cause the connection to get disconnected.
1230 Daniel Stenberg (26 Jul 2009)
1231 - Johan van Selst posted bug report #2825989
1232 (http://curl.haxx.se/bug/view.cgi?id=2825989) pointing out that
1233 OpenSSL-powered libcurl didn't support the SHA-2 digest algorithm, and
1234 provided the solution too: to use OpenSSL_add_all_algorithms() in addition
1235 to the older SSLeay_* alternative. OpenSSL_add_all_algorithms was added in
1238 Daniel Stenberg (23 Jul 2009)
1239 - Added CURLOPT_SSH_KNOWNHOSTS, CURLOPT_SSH_KEYFUNCTION, CURLOPT_SSH_KEYDATA.
1240 They introduce known_host support for SSH keys to libcurl. See docs for
1241 details. Note that this feature depends on a new enough libssh2 version, to
1242 be supported in libssh2 1.2 and later (or current git repo at this time).
1244 Michal Marek (22 Jul 2009)
1245 - David Binderman found a memory and fd leak in lib/gtls.c:load_file()
1246 (https://bugzilla.novell.com/523919). When looking at the code, I found that
1247 also the ptr pointer can leak.
1249 Kamil Dudka (20 Jul 2009)
1250 - Claes Jakobsson improved the support for client certificates handling in
1251 NSS-powered libcurl. Now the client certificates can be selected
1252 automatically by a NSS built-in hook. Additionally pre-login to all PKCS11
1253 slots is no more performed. It used to cause problems with HW tokens.
1255 - Fixed reference counting for NSS client certificates. Now the PEM reader
1256 module should be always properly unloaded on Curl_nss_cleanup(). If the
1257 unload fails though, libcurl will try to reuse the already loaded instance.
1259 Daniel Fandrich (15 Jul 2009)
1260 - Added nonblock.c to the non-automake makefiles (note that the dependencies
1261 in the Watcom makefiles aren't quite correct).
1263 Michal Marek (15 Jul 2009)
1264 - Changed the description of CURLINFO_OS_ERRNO to make it clear that the
1265 errno is not reset on success.
1267 Guenter Knauf (14 Jul 2009)
1268 - renamed generated config.h to curl_config.h to avoid any future clashes
1269 with config.h from other projects.
1271 Daniel Stenberg (9 Jul 2009)
1272 - Eric Wong introduced curlx_nonblock() that the curl tool now (re-)uses for
1273 setting a file descriptor non-blocking. Used by the functionality Eric
1274 himself brough on June 15th.
1276 Daniel Stenberg (8 Jul 2009)
1277 - Constantine Sapuntzakis posted bug report #2813123
1278 (http://curl.haxx.se/bug/view.cgi?id=2813123) and an a patch that fixes the
1281 Url A is accessed using auth. Url A redirects to Url B (on a different
1282 server0. Url B reuses a persistent connection. Url B has auth, even though
1283 it's on a different server.
1285 Note: if Url B does not reuse a persistent connection, auth is not sent.
1289 data->state.first_host is not initialized becuase Curl_http_connect is not
1290 called when a connection is reused.
1294 move initialization of data->state.first_host to Curl_http. No code before
1295 Curl_http uses data->state.first_host anyway.
1297 Guenter Knauf (4 Jul 2009)
1298 - Markus Koetter provided a patch to avoid getnameinfo() usage which broke a
1299 couple of both IPv4 and IPv6 autobuilds.
1301 Daniel Stenberg (29 Jun 2009)
1302 - Markus Koetter made CURLOPT_FTPPORT (and curl's -P/--ftpport) support a port
1303 range if given colon-separated after the host name/address part. Like
1304 "192.168.0.1:2000-10000"
1306 - Modified the separators used for CURLOPT_CERTINFO in multi-part outputs. I
1307 don't know how they got wrong in the first place, but using this output
1308 format makes it possible to quite easily separate the string into an array
1311 Daniel Fandrich (16 June 2009)
1312 - Added a few more compiler warning options for gcc.
1314 Daniel Stenberg (16 Jun 2009)
1315 - Reuven Wachtfogel made curl -o - properly produce a binary output on windows
1316 (no newline translations). Use -B/--use-ascii if you rather get the ascii
1319 Michal Marek (16 Jun 2009)
1320 - When doing non-anonymous ftp via http proxies and the password is not
1321 provided in the url, add it there (squid needs this).
1323 Daniel Stenberg (15 Jun 2009)
1324 - Eric Wong's patch:
1326 This allows curl(1) to be used as a client-side tunnel for arbitrary stream
1327 protocols by abusing chunked transfer encoding in both the HTTP request and
1328 HTTP response. This requires server support for sending a response while a
1329 request is still being read, of course.
1331 If attempting to read from stdin returns EAGAIN, then we pause our sender.
1332 This leaves curl to attempt to read from the socket while reading from stdin
1333 (and thus sending) is paused.
1335 This change was needed to allow successfully tunneling the git protocol over
1336 HTTP (--no-buffer is needed, as well).
1338 Patrick Monnerat (15 Jun 2009)
1339 - Replaced use of standard C library rand()/srand() by our own pseudo-random
1342 Yang Tse (11 Jun 2009)
1343 - I adapted testcurl script to allow building test harness programs when
1344 cross-compiling for a *-*-mingw* host.
1346 Daniel Stenberg (10 Jun 2009)
1347 - Fabian Keil ran clang on the (lib)curl code, found a bunch of warnings and
1348 contributed a range of patches to fix them.
1350 Yang Tse (10 Jun 2009)
1351 - I introduced configure script option --enable-curldebug which now allows
1352 the decoupled enabling or disabling of the curl debug memory tracking
1353 feature from the --enable-debug option which no longer controls this.
1355 curl --version will list 'Debug' feature for debug enabled builds, and
1356 will list 'TrackMemory' feature for curl debug memory tracking capable
1357 builds. These features are independent and can be controlled when running
1358 the configure script. When --enable-debug is given both features will be
1359 enabled, unless some restriction prevents memory tracking from being used.
1361 Internally, definition of preprocessor symbol DEBUGBUILD restricts code
1362 which is only compiled for debug enabled builds. And symbol CURLDEBUG is
1363 used to differentiate code which is _only_ used for memory tracking.
1365 Yang Tse (9 Jun 2009)
1366 - Daniel Steinberg pointed out that Curl_FormInit() in formdata.c was not
1367 initializing the fread callback pointer and this triggered a compiler
1368 warning, also provided a friendly suggestion on how to fix it.
1370 Daniel Stenberg (8 Jun 2009)
1371 - Claes Jakobsson provided a patch for libcurl-NSS that fixed a bad refcount
1372 issue with client certs that caused issues like segfaults.
1373 http://curl.haxx.se/mail/lib-2009-05/0316.html
1375 - Triggered by bug report #2798852 and the patch in there, I fixed configure
1376 to detect gnutls build options with pkg-config only and not libgnutls-config
1377 anymore since GnuTLS has stopped distributing that tool. If an explicit path
1378 is given to configure, we will instead guess on how to link and use that
1379 lib. I did not use the patch from the bug report.
1381 Yang Tse (8 Jun 2009)
1382 - Igor Novoseltsev adjusted Makefile.vxworks to get sources and headers
1383 included from Makefile.inc, and provided docs\INSTALL VxWorks section.
1385 - I removed buildconf.bat from release and daily snapshot archives. This
1386 file is only for CVS tree checkout builds.
1388 Daniel Stenberg (8 Jun 2009)
1389 - Eric Wong fixed --no-buffer to actually switch off output buffering. Been
1392 Bill Hoffman (6 Jun 2009)
1393 - Added some cmake docs and fixed socklen_t in the build.
1395 Yang Tse (5 Jun 2009)
1396 - John E. Malmberg provided VMS specific patch: "This fixes an existing bug
1397 in urlglob.c where it was not converting the Curl Unix exit code to a VMS
1398 DCL compatible exit code. This fix required the enhancement described next.
1399 This also adds an enhancement to main.c so that when curl is run under a
1400 Unix shell like Bash on VMS, it will return the standard Unix exit codes
1401 and messages." And another patch for docs/examples.
1403 I introduced os-specific.c and os-specific.h for use in curl tool code
1404 and adjusted John E. Malmberg's patch placement to use these new files
1405 as an effort to prevent main.c from growing ad infinitum. Code already
1406 existing in main.c which is OS specific should be moved into these files.
1408 Daniel Stenberg (4 June 2009)
1409 - Setting the Content-Length: header from your app when you do a POST or PUT
1410 is almost always a VERY BAD IDEA. Yet there are still apps out there doing
1411 this, and now recently it triggered a bug/side-effect in libcurl as when
1412 libcurl sends a POST or PUT with NTLM, it sends an empty post first when it
1413 knows it will just get a 401/407 back. If the app then replaced the
1414 Content-Length header, it caused the server to wait for input that libcurl
1415 wouldn't send. Aaron Oneal reported this problem in bug report #2799008
1416 (http://curl.haxx.se/bug/view.cgi?id=2799008) and helped us verify the fix.
1418 Yang Tse (4 Jun 2009)
1419 - Igor Novoseltsev provided patches and information, that after some
1420 adjustments to better fit curl's way of doing things, have resulted
1421 in the posibility of building libcurl for VxWorks.
1423 Daniel Fandrich (2 June 2009)
1424 - Checked in a Google Android make file. To use it, you must first
1425 create a config.h file by running configure in the Android environment,
1426 which doesn't seem to be easy to do. If no easy way can be found, a
1427 static config-android.h may need to be created and checked in to the
1428 libcurl source tree.
1430 Daniel Stenberg (1 June 2009)
1431 - Claes Jakobsson fixed the configure script to better find and use NSS
1434 Yang Tse (1 Jun 2009)
1435 - John E. Malmberg provided a VMS specific clean-up for curl.h, and pointed
1436 out that the configure script was failing to detect the timeval struct on
1437 VMS when building with _XOPEN_SOURCE_EXTENDED undefined due to definition
1438 taking place in socket.h instead of time.h. I have adjusted configure
1439 script to also include this header when checking struct timeval.
1441 Daniel Stenberg (27 May 2009)
1442 - Frank McGeough provided a small OpenSSL #include fix to make libcurl compile
1443 fine with Nokia 5th edition 1.0 SDK for Symbian.
1445 - Andre Guibert de Bruet found a call to a OpenSSL function that didn't check
1446 for a failure properly.
1448 - Mike Crowe pointed out that setting CURLOPT_USERPWD to NULL used to clear
1449 the auth credentials back in 7.19.0 and earlier while now you have to set ""
1450 to get the same effect. His patch brings back the ability to use NULL.
1452 - Claes Jakobsson fixed libcurl-NSS to build fine even without the
1453 PK11_CreateGenericObject() function.
1455 Daniel Stenberg (25 May 2009)
1456 - bug report #2796358 (http://curl.haxx.se/bug/view.cgi?id=2796358) pointed
1457 out that the cookie parser would leak memory when it parses cookies that are
1458 received with domain, path etc set multiple times in the same header. While
1459 such a cookie is questionable, they occur in the wild and libcurl no longer
1460 leaks memory for them. I added such a header to test case 8.
1462 Daniel Fandrich (22 May 2009)
1463 - Removed some obsolete digest code that caused a valgrind error in test 551.
1465 Daniel Fandrich (20 May 2009)
1466 - Added "non-existing host" test keywords to make it easy to skip those
1467 tests on machines that have broken DNS configurations (such as
1468 those configured to use OpenDNS).
1470 Daniel Stenberg (19 May 2009)
1471 - Kamil Dudka brought the patch from the Redhat bug entry
1472 https://bugzilla.redhat.com/show_bug.cgi?id=427966 which was libcurl closing
1473 a bad file descriptor when closing down the FTP data connection. Caolan
1474 McNamara seems to be the original author of it.
1476 Version 7.19.5 (18 May 2009)
1478 Daniel Stenberg (17 May 2009)
1479 - James Bursa posted a patch to the mailing list that fixed a problem with
1480 no_proxy which made it not skip the proxy if the URL entered contained a
1481 user name. I added test case 1101 to verify.
1483 Daniel Stenberg (11 May 2009)
1484 - Balint Szilakszi reported a memory leak when libcurl did gzip decompression
1485 of streams that had some parts (legitimately) missing. We now provide and use
1486 a proper cleanup function for the content encoding submodule.
1487 http://curl.haxx.se/mail/lib-2009-05/0092.html
1489 - Kamil Dudka provided a fix for libcurl-NSS reported by Michael Cronenworth
1490 at https://bugzilla.redhat.com/show_bug.cgi?id=453612#c12
1492 If an incorrect password is given while loading a private key, libcurl ends
1493 up in an infinite loop consuming memory. The bug is critical.
1495 - I fixed the problem with doing NTLM, POST and then following a 302 redirect,
1496 as reported by Ebenezer Ikonne (on curl-users) and Laurent Rabret (on
1497 curl-library). The transfer was mistakenly marked to get more data to send
1498 but since it didn't actually have that, it just hung there...
1500 Daniel Stenberg (10 May 2009)
1501 - Andre Guibert de Bruet correctly pointed out an over-alloc with one wasted
1502 byte in the digest code.
1504 Yang Tse (9 May 2009)
1505 - Removed DOS and TPF package's subdirectory Makefile.am, it was only used
1506 to include some files in the distribution tarball serving no other purpose.
1507 Files from the DOS and TPF subdirectories are now included in the EXTRA_DIST
1508 of the Makefile in the parent subdirectory.
1510 Yang Tse (8 May 2009)
1511 - Changed host name literal in several tests to one under the haxx.se domain.
1513 - Renamed vc6 workspace and project files to avoid filename clash when used
1514 for conversion to later VS versions.
1516 Daniel Stenberg (8 May 2009)
1517 - Constantine Sapuntzakis fixed bug report #2784055
1518 (http://curl.haxx.se/bug/view.cgi?id=2784055) identifying a problem to
1519 connect to SOCKS proxies when using the multi interface. It turned out to
1520 almost not work at all previously. We need to wait for the TCP connect to
1521 be properly verified before doing the SOCKS magic.
1523 There's still a flaw in the FTP code for this.
1525 Daniel Stenberg (7 May 2009)
1526 - Made the SO_SNDBUF setting for the data connection socket for ftp uploads as
1527 well. See change 28 Apr 2009.
1529 Yang Tse (7 May 2009)
1530 - Fixed an issue affecting FTP transfers, introduced with the transfer.c
1531 patch committed May 4.
1533 Daniel Stenberg (7 May 2009)
1534 - Man page *roff problems fixed thanks to input from Colin Watson. Problems
1535 reported in the Debian package.
1537 - Vijay G filed bug report #2723236
1538 (http://curl.haxx.se/bug/view.cgi?id=2723236) identifying a problem with
1539 libcurl's TFTP code and its lack of dealing with the OACK packet.
1541 Yang Tse (5 May 2009)
1542 - Fixed the --ftp-port address of test #251 to the CLIENTIP address, and
1543 reverted the change affecting test suite harness committed 4 May.
1545 Daniel Stenberg (5 May 2009)
1546 - Inspired by Michael Smith's session id fix for OpenSSL, I did the
1547 corresponding fix in the GnuTLS code: make sure to store the new session id
1548 in case the previous re-used one is rejected.
1550 Daniel Stenberg (4 May 2009)
1551 - Michael Smith posted bug report #2786255
1552 (http://curl.haxx.se/bug/view.cgi?id=2786255) with a patch, identifying how
1553 libcurl did not deal with SSL session ids properly if the server rejected a
1554 re-use of one. Starting now, it will forget the rejected one and remember
1555 the new. This change was for OpenSSL only, it is likely that other SSL lib
1556 code needs similar fixes.
1558 Yang Tse (4 May 2009)
1559 - Applied David McCreedy's "transfer.c fixes for CURL_DO_LINEEND_CONV and
1560 non-ASCII platform HTTP requests" patch addressing two HTTP PUT problems:
1561 1) On non-ASCII platforms not all of the protocol portions of the PUT are
1562 being translated to ASCII. 2) On all platforms the line endings of part of
1563 the protocol portions are mangled from CRLF to CRCRLF if data->set.crlf or
1564 data->set.prefer_ascii are set (depending on CURL_DO_LINEEND_CONV).
1566 - Applied David McCreedy's patch to fix test suite harness to allow test FTP
1567 server and client on different machines, providing FTP client address when
1568 running the FTP test server.
1570 Daniel Fandrich (3 May 2009)
1571 - Added and disabled test case 563 which shows KNOWN_BUGS #59. The bug
1572 report failed to mention that a proxy must be used to reproduce it.
1574 Yang Tse (2 May 2009)
1575 - Use a build-time configured curl_socklen_t data type instead of socklen_t.
1577 Yang Tse (1 May 2009)
1578 - Applied David McCreedy's patches "TPF-platform specific changes to various
1579 files" and "http.c fix to Curl_proxyCONNECT for non-ASCII platforms", the
1580 former with minor edits.
1582 Daniel Stenberg (30 Apr 2009)
1583 - I was going to fix issue #59 in KNOWN_BUGS
1585 If the CURLOPT_PORT option is used on an FTP URL like
1586 "ftp://example.com/file;type=A" the ";type=A" is stripped off.
1588 I added test case 562 to verify, only to find out that I couldn't repeat
1589 this bug so I hereby consider it not a bug anymore!
1591 Daniel Stenberg (29 Apr 2009)
1592 - Based on bug report #2723219 (http://curl.haxx.se/bug/view.cgi?id=2723219)
1593 I've now made TFTP "connections" not being kept for re-use within libcurl.
1594 TFTP is UDP-based so the benefit was really low (if even existing) to begin
1595 with so instead of tracking down to fix this problem we instead removed the
1596 re-use. I also enabled test case 1099 that I wrote a few days ago to verify
1597 that this change fixes the reported problem.
1599 Daniel Stenberg (28 Apr 2009)
1600 - Constantine Sapuntzakis filed bug report #2783090
1601 (http://curl.haxx.se/bug/view.cgi?id=2783090) pointing out that on windows
1602 we need to grow the SO_SNDBUF buffer somewhat to get really good upload
1603 speeds. http://support.microsoft.com/kb/823764 has the details. Friends
1604 confirmed that simply adding 32 to CURL_MAX_WRITE_SIZE is enough.
1606 - Bug report #2709004 (http://curl.haxx.se/bug/view.cgi?id=2709004) by Tim
1607 Chen pointed out how curl couldn't upload with resume when reading from a
1610 This ended up with the introduction of a new return code for the
1611 CURLOPT_SEEKFUNCTION callback that basically says that the seek failed but
1612 that libcurl may try to resolve the situation anyway. In our case this means
1613 libcurl will attempt to instead read that much data from the stream instead
1614 of seeking and that way curl can now upload with resume when data is read
1617 Daniel Stenberg (26 Apr 2009)
1618 - Bug report #2779733 (http://curl.haxx.se/bug/view.cgi?id=2779733) by Sven
1619 Wegener pointed out that CURLINFO_APPCONNECT_TIME didn't work with the multi
1620 interface and provided a patch that fixed the problem!
1622 Daniel Stenberg (24 Apr 2009)
1623 - Kamil Dudka fixed another NSS-related leak when client certs were used.
1625 - Bug report #2779245 (http://curl.haxx.se/bug/view.cgi?id=2779245) by Rainer
1626 Koenig pointed out that the man page didn't tell that the *_proxy
1627 environment variables can be specified lower case or UPPER CASE and the
1628 lower case takes precedence,
1630 Daniel Fandrich (21 Apr 2009)
1631 - Added new libcurl source files to Amiga, RiscOS and VC6 build files.
1633 Yang Tse (21 Apr 2009)
1634 - Moved potential inclusion of system's malloc.h and memory.h header files to
1635 setup_once.h. Inclusion of each header file is based on the definition of
1636 NEED_MALLOC_H and NEED_MEMORY_H respectively.
1638 Renamed libcurl's memory.h to curl_memory.h
1640 Daniel Stenberg (20 Apr 2009)
1641 - Leanic Lefever reported a crash and did some detailed research on why and
1642 how it occurs (http://curl.haxx.se/mail/lib-2009-04/0289.html). The
1643 conclusion was that if an error is detected and Curl_done() is called for
1644 the connection, ftp_done() could at times return another error code that
1645 then would take precedence and that new code confused existing logic that
1646 works for the first error code (CURLE_SEND_ERROR) only.
1648 - Gisle Vanem noticed that --libtool would produce bogus strings at times for
1649 OBJECTPOINT options. Now we've introduced a new function - my_setopt_str -
1650 within the app for setting plain string options to avoid the risk of this
1653 Daniel Stenberg (17 Apr 2009)
1654 - Pramod Sharma reported and tracked down a bug when doing FTP over a HTTP
1655 proxy. libcurl would then wrongly close the connection after each
1656 request. In his case it had the weird side-effect that it killed NTLM auth
1657 for the proxy causing an inifinite loop!
1659 I added test case 1098 to verify this fix. The test case does however not
1660 properly verify that the transfers are done persistently - as I couldn't
1661 think of a clever way to achieve it right now - but you need to read the
1662 stderr output after a test run to see that it truly did the right thing.
1664 Daniel Stenberg (13 Apr 2009)
1665 - bug report #2727981 (http://curl.haxx.se/bug/view.cgi?id=2727981) by Martin
1666 Storsjö pointed out how setting CURLOPT_NOBODY to 0 could be downright
1667 confusing as it set the method to either GET or HEAD. The example he showed
1670 curl_easy_setopt(curl, CURLOPT_PUT, 1);
1671 curl_easy_setopt(curl, CURLOPT_NOBODY, 0);
1673 The new way doesn't alter the method until the request is about to start. If
1674 CURLOPT_NOBODY is then 1 the HTTP request will be HEAD. If CURLOPT_NOBODY is
1675 0 and the request happens to have been set to HEAD, it will then instead be
1676 set to GET. I believe this will be less surprising to users, and hopefully
1677 not hit any existing users badly.
1679 - Toshio Kuratomi reported a memory leak problem with libcurl+NSS that turned
1680 out to be leaking cacerts. Kamil Dudka helped me complete the fix. The issue
1681 is found in Redhat's bug tracker:
1682 https://bugzilla.redhat.com/show_bug.cgi?id=453612
1684 There are still memory leaks present, but they seem to have other reasons.
1686 Daniel Fandrich (11 Apr 2009)
1687 - Added new libcurl source files to Symbian OS build files.
1688 - Improved Symbian support for SSL.
1690 Yang Tse (10 Apr 2009)
1691 - Daniel Johnson improved the MacOSX-Framework shell script to now perform all
1692 the steps required to build a Mac OS X four way fat ppc/i386/ppc64/x86_64
1693 libcurl.framework. Four way fat framework requires OS X 10.5 SDK or later.
1695 Yang Tse (8 Apr 2009)
1696 - Removed Sun compilers preprocessor block from curlbuild.h.dist, this also
1697 removes it from the curlbuild.h file originally distributed by the cURL
1698 project as this file is intended for systems not capable of running the
1699 configure script. For those who have been building curl out of the source
1700 code curl distribution tarball provided by curl.haxx.se the change implies
1701 nothing. Previous change in this area committed 2 Apr becomes irrelevant.
1703 Daniel Stenberg (6 Apr 2009)
1704 - I clarified in the docs that CURLOPT_SEEKFUNCTION should return 0 on success
1705 and 1 on fatal errors. Previously it only mentioned non-zero on fatal
1706 errors. This is a slight change in meaning, but it follows what we've done
1707 elsewhere before and it opens up for LOTS of more useful return codes
1708 whenever we can think of them...
1710 Yang Tse (2 Apr 2009)
1711 - Fix curl_off_t definition for builds done using Sun compilers and a
1712 non-configured libcurl. In this case curl_off_t data type was gated
1713 to the off_t data type which depends on the _FILE_OFFSET_BITS. This
1714 configuration is exactly the unwanted configuration for our curl_off_t
1715 data type which must not depend on such setting. This breaks ABI for
1716 libcurl libraries built with Sun compilers which were built without
1717 having run the configure script with _FILE_OFFSET_BITS different than
1718 64 and using the ILP32 data model.
1720 Daniel Stenberg (1 Apr 2009)
1721 - Andre Guibert de Bruet fixed a NULL pointer use in an infof() call if a
1722 strdup() call failed.
1724 Daniel Fandrich (31 Mar 2009)
1725 - Properly return an error code in curl_easy_recv (reported by Jim Freeman).
1727 Daniel Stenberg (18 Mar 2009)
1728 - Kamil Dudka brought a patch that enables 6 additional crypto algorithms when
1729 NSS is used. These ciphers were added in NSS 3.4 and require to be enabled
1732 Daniel Stenberg (13 Mar 2009)
1733 - Use libssh2_version() to present the libssh2 version in case the libssh2
1734 library is found to support it.
1736 Yang Tse (12 Mar 2009)
1737 - Added missing Curl_read() return code checking in TELNET transfers.
1739 - Pierre Brico found and fixed TELNET transfers not being aborted upon
1740 a write callback failure.
1742 Daniel Stenberg (11 Mar 2009)
1743 - Kamil Dudka made the curl tool properly call curl_global_init() before any
1744 other libcurl function.
1746 Yang Tse (11 Mar 2009)
1747 - Added missing TELNET timeout support for Windows builds. This issue was
1748 reported by Pierre Brico.
1750 Daniel Stenberg (9 Mar 2009)
1751 - Frank Hempel found out a bug and provided the fix:
1753 curl_easy_duphandle did not necessarily duplicate the CURLOPT_COOKIEFILE
1754 option. It only enabled the cookie engine in the destination handle if
1755 data->cookies is not NULL (where data is the source handle). In case of a
1756 newly initialized handle which just had the cookie support enabled by a
1757 curl_easy_setopt(handle, CURL_COOKIEFILE, "")-call, handle->cookies was
1758 still NULL because the setopt-call only appends the value to
1759 data->change.cookielist, hence duplicating this handle would not have the
1760 cookie engine switched on.
1762 We also concluded that the slist-functionality would be suitable for being
1763 put in its own module rather than simply hanging out in lib/sendf.c so I
1764 created lib/slist.[ch] for them.
1766 - Andreas Farber made the 'buildconf' script check for the presence of m4
1767 scripts to make it detect a bad checkout earlier. People with older
1768 checkouts who don't do cvs update with the -d option won't get the new dirs
1769 and then will get funny outputs that can be a bit hard to understand and
1772 Daniel Stenberg (8 Mar 2009)
1773 - Andre Guibert de Bruet found and fixed a code segment in ssluse.c where the
1774 allocation of the memory BIO was not being properly checked.
1776 - Andre Guibert de Bruet fixed the gnutls-using code: There are a few places
1777 in the gnutls code where we were checking for negative values for errors,
1778 when the man pages state that GNUTLS_E_SUCCESS is returned on success and
1779 other values indicate error conditions.
1781 - Bill Egert pointed out (http://curl.haxx.se/bug/view.cgi?id=2671602) that
1782 curl didn't use sprintf() in a way that is documented to work in POSIX but
1783 since we use our own printf() code (from libcurl) that shouldn't be a
1784 problem. Nonetheless I modified the code to not rely on such particular
1785 features and to not cause further raised eyebrowse with no good reason.
1787 Daniel Fandrich (5 Mar 2009)
1788 - Expanded the security section of the libcurl-tutorial man page to cover
1789 more issues for authors to consider when writing robust libcurl-using
1792 Yang Tse (5 Mar 2009)
1793 - Fixed NTLM authentication memory leak on SSPI enabled Windows builds. This
1794 issue was noticed by Chris Deidun.
1796 Daniel Fandrich (4 Mar 2009)
1797 - Fixed a problem with m4 quoting in the OpenSSL configure check reported
1800 Daniel Stenberg (3 Mar 2009)
1801 - David James brought a patch that make libcurl close (all) dead connections
1802 whenever you attempt to open a new connection.
1804 1. After cleaning up a dead connection, "continue" instead of
1805 returning FALSE. This ensures that we clean up all dead connections,
1806 rather than just cleaning up the first dead connection.
1807 2. Move up the cleanup for dead connections so that it occurs for
1808 all connections, rather than just the connections which have the same
1809 preferences as our current new connection.
1811 Version 7.19.4 (3 March 2009)
1813 Daniel Stenberg (3 Mar 2009)
1814 - David Kierznowski notified us about a security flaw
1815 (http://curl.haxx.se/docs/adv_20090303.html also known as CVE-2009-0037) in
1816 which previous libcurl versions (by design) can be tricked to access an
1817 arbitrary local/different file instead of a remote one when
1818 CURLOPT_FOLLOWLOCATION is enabled. This flaw is now fixed in this release
1819 together this the addition of two new setopt options for controlling this
1822 o CURLOPT_REDIR_PROTOCOLS controls what protocols libcurl is allowed to
1823 follow to when CURLOPT_FOLLOWLOCATION is enabled. By default, this option
1824 excludes the FILE and SCP protocols and thus you nee to explicitly allow
1825 them in your app if you really want that behavior.
1827 o CURLOPT_PROTOCOLS controls what protocol(s) libcurl is allowed to fetch
1828 using the primary URL option. This is useful if you want to allow a user or
1829 other outsiders control what URL to pass to libcurl and yet not allow all
1830 protocols libcurl may have been built to support.
1832 Daniel Stenberg (27 Feb 2009)
1833 - Senthil Raja Velu reported a problem when CURLOPT_INTERFACE and
1834 CURLOPT_LOCALPORT were used together (the local port bind failed), and
1835 Markus Koetter provided the fix!
1837 Daniel Stenberg (25 Feb 2009)
1838 - As Daniel Fandrich figured out, we must do the GnuTLS initing in the
1839 curl_global_init() function to properly maintain the performing functions
1840 thread-safe. We've previously (28 April 2007) moved the init to a later time
1841 just to avoid it to fail very early when libgcrypt dislikes the situation,
1842 but that move was bad and the fix should rather be in libgcrypt or
1845 Daniel Stenberg (24 Feb 2009)
1846 - Brian J. Murrell found out that Negotiate proxy authentication didn't work.
1847 It happened because the code used the struct for server-based auth all the
1848 time for both proxy and server auth which of course was wrong.
1850 Daniel Stenberg (23 Feb 2009)
1851 - After a bug reported by James Cheng I've made curl_easy_getinfo() for
1852 CURLINFO_CONTENT_LENGTH_DOWNLOAD and CURLINFO_CONTENT_LENGTH_UPLOAD return
1853 -1 if the sizes aren't know. Previously these returned 0, make it impossible
1854 to detect the difference between actually zero and unknown.
1856 Yang Tse (23 Feb 2009)
1857 - Daniel Johnson provided a shell script that will perform all the steps needed
1858 to build a Mac OS X fat ppc/i386 or ppc64/x86_64 libcurl.framework
1860 Daniel Stenberg (23 Feb 2009)
1861 - I renamed everything in the windows builds files that used the name 'curllib'
1862 to the proper 'libcurl' as clearly this caused confusion.
1864 Yang Tse (20 Feb 2009)
1865 - Do not halt compilation when using VS2008 to build a Windows 2000 target.
1867 Daniel Stenberg (20 Feb 2009)
1868 - Linus Nielsen Feltzing reported and helped me repeat and fix a problem with
1869 FTP with the multi interface: when a transfer fails, like when aborted by a
1870 write callback, the control connection was wrongly closed and thus not
1873 This change is also an attempt to cleanup the code somewhat in this area, as
1874 now the FTP code attempts to keep (better) track on pending responses
1875 necessary to get read in ftp_done().
1877 Daniel Stenberg (19 Feb 2009)
1878 - Patrik Thunstrom reported a problem and helped me repeat it. It turned out
1879 libcurl did a superfluous 1000ms wait when doing SFTP downloads!
1881 We read data with libssh2 while doing the "DO" operation for SFTP and then
1882 when we were about to start getting data for the actual file part, the
1883 "TRANSFER" part, we waited for socket action (in 1000ms) before doing a
1884 libssh2-read. But in this case libssh2 had already read and buffered the
1885 data so we ended up always just waiting 1000ms before we get working on the
1888 Patrick Monnerat (18 Feb 2009)
1889 - FTP downloads (i.e.: RETR) ending with code 550 now return error
1890 CURLE_REMOTE_FILE_NOT_FOUND instead of CURLE_FTP_COULDNT_RETR_FILE.
1892 Daniel Stenberg (17 Feb 2009)
1893 - Kamil Dudka made NSS-powered builds compile and run again!
1895 - A second follow-up change by Andre Guibert de Bruet to fix a related memory
1896 leak like that fixed on the 14th. When zlib returns failure, we need to
1897 cleanup properly before returning error.
1899 - CURLOPT_FTP_CREATE_MISSING_DIRS can now be set to 2 in addition to 1 for
1900 plain FTP connections, and it will then allow MKD to fail once and retry the
1901 CWD afterwards. This is especially useful if you're doing many simultanoes
1902 connections against the same server and they all have this option enabled,
1903 as then CWD may first fail but then another connection does MKD before this
1904 connection and thus MKD fails but trying CWD works! The numbers can
1905 (should?) now be set with the convenience enums now called
1906 CURLFTP_CREATE_DIR and CURLFTP_CREATE_DIR_RETRY.
1908 Tests has proven that if you're making an application that uploads a set of
1909 files to an ftp server, you will get a noticable gain in speed if you're
1910 using multiple connections and this option will be then be very useful.
1912 Daniel Stenberg (14 Feb 2009)
1913 - Andre Guibert de Bruet found and fixed a memory leak in the content encoding
1914 code, which could happen on libz errors.
1916 Daniel Fandrich (12 Feb 2009)
1917 - Added support for Digest and NTLM authentication using GnuTLS.
1919 Daniel Stenberg (11 Feb 2009)
1920 - CURLINFO_CONDITION_UNMET was added to allow an application to get to know if
1921 the condition in the previous request was unmet. This is typically a time
1922 condition set with CURLOPT_TIMECONDITION and was previously not possible to
1923 reliably figure out. From bug report #2565128
1924 (http://curl.haxx.se/bug/view.cgi?id=2565128) filed by Jocelyn Jaubert.
1926 Daniel Fandrich (4 Feb 2009)
1927 - Don't add the standard /usr/lib or /usr/include paths to LDFLAGS and CPPFLAGS
1928 (respectively) when --with-ssl=/usr is used (patch based on FreeBSD).
1930 - Added an explicit buffer limit check in msdosify() (patch based on FreeBSD).
1931 This couldn't ever overflow in curl, but might if the code were used
1932 elsewhere or under different conditions.
1934 Daniel Stenberg (3 Feb 2009)
1935 - Hidemoto Nakada provided a small fix that makes it possible to get the
1936 CURLINFO_CONTENT_LENGTH_DOWNLOAD size from file:// "transfers" with
1937 CURLOPT_NOBODY set true.
1939 Daniel Stenberg (2 Feb 2009)
1940 - Patrick Scott found a rather large memory leak when using the multi
1941 interface and setting CURLMOPT_MAXCONNECTS to something less than the number
1942 of handles you add to the multi handle. All the connections that didn't fit
1943 in the cache would not be properly disconnected nor freed!
1945 - Craig A West brought us: libcurl now defaults to do CONNECT with HTTP
1946 version 1.1 instead of 1.0 like before. This change also introduces the new
1947 proxy type for libcurl called 'CURLPROXY_HTTP_1_0' that then allows apps to
1948 switch (back) to CONNECT 1.0 requests. The curl tool also got a --proxy1.0
1949 option that works exactly like --proxy but sets CURLPROXY_HTTP_1_0.
1951 I updated all test cases cases that use CONNECT and I tried to do some using
1952 --proxy1.0 and some updated to do CONNECT 1.1 to get both versions run.
1954 Daniel Stenberg (31 Jan 2009)
1955 - When building with c-ares 1.6.1 (not yet released) or later and IPv6 support
1956 enabled, we can now take advantage of its brand new AF_UNSPEC support in
1957 ares_gethostbyname(). This makes test case 241 finally run fine for me with
1958 this setup since it now parses the "::1 ip6-localhost" line fine in my
1961 Daniel Stenberg (30 Jan 2009)
1962 - Scott Cantor filed bug report #2550061
1963 (http://curl.haxx.se/bug/view.cgi?id=2550061) mentioning that I failed to
1964 properly make sure that the VC9 makefiles got included in the latest
1965 release. I've now fixed the release script and verified it so next release
1966 will hopefully include them properly!
1968 Daniel Fandrich (30 Jan 2009)
1969 - Fixed --disable-proxy for FTP and SOCKS. Thanks to Daniel Egger for
1972 Yang Tse (29 Jan 2009)
1973 - Introduced curl_sspi.c and curl_sspi.h for the implementation of functions
1974 Curl_sspi_global_init() and Curl_sspi_global_cleanup() which previously were
1975 named Curl_ntlm_global_init() and Curl_ntlm_global_cleanup() in http_ntlm.c
1976 Also adjusted socks_sspi.c to remove the link-time dependency on the Windows
1977 SSPI library using it now in the same way as it was done in http_ntlm.c.
1979 Daniel Stenberg (28 Jan 2009)
1980 - Markus Moeller introduced two new options to libcurl:
1981 CURLOPT_SOCKS5_GSSAPI_SERVICE and CURLOPT_SOCKS5_GSSAPI_NEC to allow libcurl
1982 to do GSS-style authentication with SOCKS5 proxies. The curl tool got the
1983 options called --socks5-gssapi-service and --socks5-gssapi-nec to enable
1986 Daniel Stenberg (26 Jan 2009)
1987 - Chad Monroe provided the new CURLOPT_TFTP_BLKSIZE option that allows an app
1988 to set desired block size to use for TFTP transfers instead of the default
1991 - The "-no_ticket" option was introduced in Openssl0.9.8j. It's a flag to
1992 disable "rfc4507bis session ticket support". rfc4507bis was later turned
1993 into the proper RFC5077 it seems: http://tools.ietf.org/html/rfc5077
1995 The enabled extension concerns the session management. I wonder how often
1996 libcurl stops a connection and then resumes a TLS session. also, sending the
1997 session data is some overhead. .I suggest that you just use your proposed
1998 patch (which explicitly disables TICKET).
2000 If someone writes an application with libcurl and openssl who wants to
2001 enable the feature, one can do this in the SSL callback.
2003 Sharad Gupta brought this to my attention. Peter Sylvester helped me decide
2004 on the proper action.
2006 - Alexey Borzov filed bug report #2535504
2007 (http://curl.haxx.se/bug/view.cgi?id=2535504) pointing out that realms with
2008 quoted quotation marks in HTTP Digest headers didn't work. I've now added
2009 test case 1095 that verifies my fix.
2011 - Craig A West brought CURLOPT_NOPROXY and the corresponding --noproxy option.
2012 They basically offer the same thing the NO_PROXY environment variable only
2013 offered previously: list a set of host names that shall not use the proxy
2014 even if one is specified.
2016 Daniel Fandrich (20 Jan 2009)
2017 - Call setlocale() for libtest tests to test the effects of locale-induced
2018 libc changes on libcurl.
2020 - Fixed a couple more locale-dependent toupper conversions, mainly for
2021 clarity. This does fix one problem that causes ;type=i FTP URLs
2022 to fail in the Turkish locale when CURLOPT_PROXY_TRANSFER_MODE is
2023 used (test case 561)
2025 - Added tests 561 and 1091 through 1094 to test various combinations
2026 of ;type= and ;mode= URLs that could potentially fail in the Turkish
2029 Daniel Stenberg (20 Jan 2009)
2030 - Lisa Xu pointed out that the ssh.obj file was missing from the
2031 lib/Makefile.vc6 file (and thus from the vc8 and vc9 ones too).
2033 Version 7.19.3 (19 January 2009)
2035 Daniel Stenberg (16 Jan 2009)
2036 - Andrew de los Reyes fixed curlbuild.h for "generic" gcc builds on PPC, both
2039 Daniel Stenberg (15 Jan 2009)
2040 - Tim Ansell fixed a compiler warning in lib/cookie.c
2042 Daniel Stenberg (14 Jan 2009)
2043 - Grant Erickson fixed timeouts for TFTP such that specifying a
2044 connect-timeout, a max-time or both options work correctly and as expected
2045 by passing the correct boolean value to Curl_timeleft via the
2046 'duringconnect' parameter.
2048 With this small change, curl TFTP now behaves as expected (and likely as
2049 originally-designed):
2051 1) For non-existent or unreachable dotted IP addresses:
2053 a) With no options, follows the default curl 300s timeout...
2054 b) With --connect-timeout only, follows that value...
2055 c) With --max-time only, follows that value...
2056 d) With both --connect-timeout and --max-time, follows the smaller value...
2058 and times out with a "curl: (7) Couldn't connect to server" error.
2060 2) For transfers to/from a valid host:
2062 a) With no options, follows default curl 300s timeout for the
2063 first XRQ/DATA/ACK transaction and the default TFTP 3600s
2064 timeout for the remainder of the transfer...
2066 b) With --connect-time only, follows that value for the
2067 first XRQ/DATA/ACK transaction and the default TFTP 3600s
2068 timeout for the remainder of the transfer...
2070 c) With --max-time only, follows that value for the first
2071 XRQ/DATA/ACK transaction and for the remainder of the
2074 d) With both --connect-timeout and --max-time, follows the former
2075 for the first XRQ/DATA/ACK transaction and the latter for the
2076 remainder of the transfer...
2078 and times out with a "curl: (28) Timeout was reached" error as
2081 Daniel Stenberg (13 Jan 2009)
2082 - Michael Wallner fixed a NULL pointer deref when calling
2083 curl_easy_setup(curl, CURLOPT_COOKIELIST, "SESS") on a CURL handle with no
2086 - Stefan Teleman brought a patch to fix the default curlbuild.h file for the
2089 Daniel Stenberg (12 Jan 2009)
2090 - Based on bug report #2498665 (http://curl.haxx.se/bug/view.cgi?id=2498665)
2091 by Daniel Black, I've now added magic to the configure script that makes it
2092 use pkg-config to detect gnutls details as well if the existing method
2093 (using libgnutls-config) fails. While doing this, I cleaned up and unified
2094 the pkg-config usage when detecting openssl and nss as well.
2096 Daniel Stenberg (11 Jan 2009)
2097 - Karl Moerder brought the patch that creates vc9 Makefiles, and I made
2098 'maketgz' now use the actual makefile targets to do the VC8 and VC9
2101 Daniel Stenberg (10 Jan 2009)
2102 - Emil Romanus fixed:
2104 When using the multi interface over HTTP and the server returns a Location
2105 header, the running easy handle will get stuck in the CURLM_STATE_PERFORM
2106 state, leaving the external event loop stuck waiting for data from the
2107 ingoing socket (when using the curl_multi_socket_action stuff). While this
2108 bug was pretty hard to find, it seems to require only a one-line fix. The
2109 break statement on line 1374 in multi.c caused the function to skip the call
2112 How to reproduce this bug? Well, that's another question. evhiperfifo.c in
2113 the examples directory chokes on this bug only _sometimes_, probably
2114 depending on how fast the URLs are added. One way of testing the bug out is
2115 writing to hiper.fifo from more than one source at the same time.
2117 Daniel Fandrich (7 Jan 2009)
2118 - Unified much of the SessionHandle initialization done in Curl_open() and
2119 curl_easy_reset() by creating Curl_init_userdefined(). This had the side
2120 effect of fixing curl_easy_reset() so it now also resets
2121 CURLOPT_FTP_FILEMETHOD and CURLOPT_SSL_SESSIONID_CACHE
2123 Daniel Stenberg (7 Jan 2009)
2124 - Rob Crittenden did once again provide an NSS update:
2126 I have to jump through a few hoops now with the NSS library initialization
2127 since another part of an application may have already initialized NSS by the
2128 time Curl gets invoked. This patch is more careful to only shutdown the NSS
2129 library if Curl did the initialization.
2131 It also adds in a bit of code to set the default ciphers if the app that
2132 call NSS_Init* did not call NSS_SetDomesticPolicy() or set specific
2133 ciphers. One might argue that this lets other application developers get
2134 lazy and/or they aren't using the NSS API correctly, and you'd be right.
2135 But still, this will avoid terribly difficult-to-trace crashes and is
2138 Daniel Stenberg (1 Jan 2009)
2139 - 'reconf' is removed since we rather have users use 'buildconf'
2141 Daniel Stenberg (31 Dec 2008)
2142 - Bas Mevissen reported http://curl.haxx.se/bug/view.cgi?id=2479030 pointing
2143 out that 'reconf' didn't properly point out the m4 subdirectory when running
2146 Daniel Stenberg (29 Dec 2008)
2147 - Phil Lisiecki filed bug report #2413067
2148 (http://curl.haxx.se/bug/view.cgi?id=2413067) that identified a problem that
2149 would cause libcurl to mark a DNS cache entry "in use" eternally if the
2150 subsequence TCP connect failed. It would thus never get pruned and refreshed
2151 as it should've been.
2153 Phil provided his own patch to this problem that while it seemed to work
2154 wasn't complete and thus I wrote my own fix to the problem.
2156 Daniel Stenberg (28 Dec 2008)
2157 - Peter Korsgaard fixed building libcurl with "configure --with-ssl
2160 - Anthony Bryan fixed more language and spelling flaws in man pages.
2162 Daniel Stenberg (22 Dec 2008)
2163 - Given a recent enough libssh2, libcurl can now seek/resume with SFTP even
2164 on file indexes beyond 2 or 4GB.
2166 - Anthony Bryan provided a set of patches that cleaned up manual language,
2167 corrected spellings and more.
2169 Daniel Stenberg (20 Dec 2008)
2170 - Igor Novoseltsev fixed a bad situation for the multi_socket() API when doing
2171 pipelining, as libcurl could then easily get confused and A) work on the
2172 handle that was not "first in queue" on a pipeline, or even B) tell the app
2173 to REMOVE a socket while it was in use by a second handle in a pipeline. Both
2174 errors caused hanging or stalling applications.
2176 Daniel Stenberg (19 Dec 2008)
2177 - curl_multi_timeout() could return a timeout value of 0 even though nothing
2178 was actually ready to get done, as the internal time resolution is higher
2179 than the returned millisecond timer. Therefore it could cause applications
2180 running on fast processors to do short bursts of busy-loops.
2181 curl_multi_timeout() will now only return 0 if the timeout is actually
2184 - Using the libssh2 0.19 function libssh2_session_block_directions(), libcurl
2185 now has an improved ability to do right when the multi interface (both
2186 "regular" and multi_socket) is used for SCP and SFTP transfers. This should
2187 result in (much) less busy-loop situations and thus less CPU usage with no
2190 Daniel Stenberg (17 Dec 2008)
2191 - SCP and SFTP with the multi interface had the same flaw: the 'DONE'
2192 operation didn't complete properly if the EAGAIN equivalent was returned but
2193 libcurl would simply continue with a half-completed close operation
2194 performed. This ruined persistent connection re-use and cause some
2195 SSH-protocol errors in general. The correction is unfortunately adding a
2196 blocking function - doing it entirely non-blocking should be considered for
2199 Gisle Vanem (16 Dec 2008)
2200 - Added the possibility to use the Watt-32 tcp/ip stack under Windows.
2201 The change simply involved adding a USE_WATT32 section in the
2202 config-win32.h files (under ./lib and ./src). This section disables
2203 the use of any Winsock headers.
2205 Daniel Stenberg (16 Dec 2008)
2206 - libssh2_sftp_last_error() was wrongly used at some places in libcurl which
2207 made libcurl sometimes not properly abort problematic SFTP transfers.
2209 Daniel Stenberg (12 Dec 2008)
2210 - More work with Igor Novoseltsev to first fix the remaining stuff for
2211 removing easy handles from multi handles when the easy handle is/was within
2212 a HTTP pipeline. His bug report #2351653
2213 (http://curl.haxx.se/bug/view.cgi?id=2351653) was also related and was
2214 eventually fixed by a patch by Igor himself.
2216 Yang Tse (12 Dec 2008)
2217 - Patrick Monnerat fixed a build regression, introduced in 7.19.2, affecting
2218 OS/400 compilations with IPv6 enabled.
2220 Daniel Stenberg (12 Dec 2008)
2221 - Mark Karpeles filed bug report #2416182 titled "crash in ConnectionExists
2222 when using duphandle+curl_mutli"
2223 (http://curl.haxx.se/bug/view.cgi?id=2416182) which showed that
2224 curl_easy_duphandle() wrongly also copied the pointer to the connection
2225 cache, which was plain wrong and caused a segfault if the handle would be
2226 used in a different multi handle than the handle it was duplicated from.
2228 Daniel Stenberg (11 Dec 2008)
2229 - Keshav Krity found out that libcurl failed to deal with dotted IPv6
2230 addresses if they were very long (>39 letters) due to a too strict address
2231 validity parser. It now accepts addresses up to 45 bytes long.
2233 Daniel Stenberg (11 Dec 2008)
2234 - Internet Explorer had a broken HTTP digest authentication before v7 and
2235 there are servers "out there" that relies on the client doing this broken
2236 Digest authentication. Apache even comes with an option to work with such
2239 The difference is only for URLs that contain a query-part (a '?'-letter and
2240 text to the right of it).
2242 libcurl now supports this quirk, and you enable it by setting the
2243 CURLAUTH_DIGEST_IE bit in the bitmask you pass to the CURLOPT_HTTPAUTH or
2244 CURLOPT_PROXYAUTH options. They are thus individually controlled to server
2247 (note that there's no way to activate this with the curl tool yet)
2249 Daniel Fandrich (9 Dec 2008)
2250 - Added test cases 1089 and 1090 to test --write-out after a redirect to
2251 test a report that the size didn't work, but these test cases pass.
2253 - Documented CURLOPT_CONNECT_ONLY as being useful only on HTTP URLs.
2255 Daniel Stenberg (9 Dec 2008)
2256 - Ken Hirsch simplified how libcurl does FTPS: now it doesn't assume any
2257 particular state for the control connection like it did before for implicit
2258 FTPS (libcurl assumed such control connections to be encrypted while some
2259 FTPS servers such as FileZilla assumes such connections to be clear
2260 mode). Use the CURLOPT_USE_SSL option to set your desired level.
2262 Daniel Stenberg (8 Dec 2008)
2263 - Fred Machado posted about a weird FTP problem on the curl-users list and when
2264 researching it, it turned out he got a 550 response back from a SIZE command
2265 and then I fell over the text in RFC3659 that says:
2267 The presence of the 550 error response to a SIZE command MUST NOT be taken
2268 by the client as an indication that the file cannot be transferred in the
2269 current MODE and TYPE.
2271 In other words: the change I did on September 30th 2008 and that has been
2272 included in the last two releases were a regression and a bad idea. We MUST
2273 NOT take a 550 response from SIZE as a hint that the file doesn't exist.
2275 - Christian Krause filed bug #2221237
2276 (http://curl.haxx.se/bug/view.cgi?id=2221237) that identified an infinite
2277 loop during GSS authentication given some specific conditions. With his
2278 patience and great feedback I managed to narrow down the problem and
2279 eventually fix it although I can't test any of this myself!
2281 Daniel Fandrich (3 Dec 2008)
2282 - Fixed the getifaddrs version of Curl_if2ip to work on systems without IPv6
2283 support (e.g. Minix)
2285 Daniel Stenberg (3 Dec 2008)
2286 - Igor Novoseltsev filed bug #2351645
2287 (http://curl.haxx.se/bug/view.cgi?id=2351645) that identified a problem with
2288 the multi interface that occured if you removed an easy handle while in
2289 progress and the handle was used in a HTTP pipeline.
2291 - Pawel Kierski pointed out a mistake in the cookie code that could lead to a
2292 bad fclose() after a fatal error had occured.
2293 (http://curl.haxx.se/bug/view.cgi?id=2382219)
2295 Daniel Fandrich (25 Nov 2008)
2296 - If a HTTP request is Basic and num is already >=1000, the HTTP test
2297 server adds 1 to num to get the data section to return. This allows
2298 testing authentication negotiations using the Basic authentication
2301 - Added tests 1087 and 1088 to test Basic authentication on a redirect
2302 with and without --location-trusted
2304 Daniel Stenberg (24 Nov 2008)
2305 - Based on a patch by Vlad Grachov, libcurl now uses a new libssh2 0.19
2306 function when built to support SCP and SFTP that helps the library to know
2307 in which direction a particular libssh2 operation would return EAGAIN so
2308 that libcurl knows what socket conditions to wait for before trying the
2309 function call again. Previously (and still when using libssh2 0.18 or
2310 earlier), libcurl will busy-loop in this situation when the easy interface
2313 Daniel Fandrich (20 Nov 2008)
2314 - Automatically detect OpenBSD's CA cert bundle.
2316 Daniel Stenberg (19 Nov 2008)
2317 - I removed the default use of "Pragma: no-cache" from libcurl when a proxy is
2318 used. It has been used since forever but it was never a good idea to use
2319 unless explicitly asked for.
2321 - Josef Wolf's extension that allows a $TESTDIR/gdbinit$testnum file that when
2322 you use runtests.pl -g, will be sourced by gdb to allow additional fancy or
2323 whatever you see fit
2325 - Christian Krause reported and fixed a memory leak that would occur with HTTP
2326 GSS/kerberos authentication (http://curl.haxx.se/bug/view.cgi?id=2284386)
2328 - Andreas Wurf and Markus Koetter helped me analyze a problem that Andreas got
2329 when uploading files to a single FTP server using multiple easy handle
2330 handles with the multi interface. Occasionally a handle would stall in
2333 The problem turned out to be a side-effect of the ConnectionExists()
2334 function's eagerness to re-use a handle for HTTP pipelining so it would
2335 select it even if already being in use, due to an inadequate check for its
2336 chances of being used for pipelnining.
2338 Daniel Fandrich (17 Nov 2008)
2339 - Added more compiler warning options for gcc 4.3
2341 Yang Tse (17 Nov 2008)
2342 - Fix a remaining problem in the inet_pton() runtime configure check. And
2343 fix internal Curl_inet_pton() failures to reject certain malformed literals.
2345 - Make configure script check if ioctl with the SIOCGIFADDR command can be
2346 used, and define HAVE_IOCTL_SIOCGIFADDR if appropriate.
2348 Daniel Stenberg (16 Nov 2008)
2349 - Christian Krause fixed a build failure when building with gss support
2350 enabled and FTP disabled.
2352 - Added check for NULL returns from strdup() in src/main.c and lib/formdata.c
2353 - reported by Jim Meyering also prevent buffer overflow on MSDOS when you do
2354 for example -O on a url with a file name part longer than PATH_MAX letters
2356 - lib/nss.c fixes based on the report by Jim Meyering: I went over and added
2357 checks for return codes for all calls to malloc and strdup that were
2358 missing. I also changed a few malloc(13) to use arrays on the stack and a
2359 few malloc(PATH_MAX) to instead use aprintf() to lower memory use.
2361 - I fixed a memory leak in Curl_nss_connect() when CURLOPT_ISSUERCERT is
2364 Daniel Fandrich (14 Nov 2008)
2365 - Added .xml as one of the few common file extensions known by the multipart
2368 - Added some #ifdefs around header files and change the EAGAIN test to
2369 fix compilation on Cell (reported by Jeff Curley).
2371 Yang Tse (14 Nov 2008)
2372 - Fixed several configure script issues affecting checks for inet_ntoa_r(),
2373 inet_ntop(), inet_pton(), getifaddrs(), fcntl() and getaddrinfo().
2375 Yang Tse (13 Nov 2008)
2376 - Refactored configure script detection of functions used to set sockets into
2377 non-blocking mode, and decouple function detection from function capability.