5 \___|\___/|_| \_\_____|
9 Daniel Stenberg (9 Apr 2010)
10 - Prefixing the FTP quote commands with an asterisk really only worked for the
11 postquote actions. This is now fixed and test case 227 has been extended to
14 Kamil Dudka (4 Apr 2010)
15 - Eliminated a race condition in Curl_resolv_timeout().
17 - Refactorized interface of Curl_ssl_recv()/Curl_ssl_send().
19 - libcurl-NSS now provides more accurate messages and error codes in case of
20 client certificate problem. Either during connection, or transfer phase.
22 Daniel Stenberg (1 Apr 2010)
23 - Matt Wixson found and fixed a bug in the SCP/SFTP area where the code
24 treated a 0 return code from libssh2 to be the same as EAGAIN while in
25 reality it isn't. The problem caused a hang in SFTP transfers from a
28 Daniel Stenberg (28 Mar 2010)
29 - Ben Greear: If you pass a URL to pop3 that does not contain a message ID as
30 part of the URL, it would previously ask for 'INBOX' which just causes the
31 pop3 server to return an error.
33 Now libcurl treats en empty message ID as a request for LIST (list of pop3
34 message IDs). User's code could then parse this and download individual
37 Daniel Stenberg (27 Mar 2010)
38 - Ben Greear brought a patch that from now on allows all protocols to specify
39 name and user within the URL, in the same manner HTTP and FTP have been
40 allowed to in the past - although far from all of the libcurl supported
41 protocls actually have that feature in their URL definition spec.
43 Daniel Stenberg (26 Mar 2010)
44 - Ben Greear brought code that makes the rate limiting code for the easy
45 interface a bit smoother as it introduces sub-second sleeps during it and it
46 also takes the buffer sizes into account.
48 Daniel Stenberg (24 Mar 2010)
49 - Bob Richmond: There's an annoying situation where libcurl will read new HTTP
50 response data from a socket, then check if it's a timeout if one is set. If
51 the last packet received constitutes the end of the response body, libcurl
52 still treats it as a timeout condition and reports a message like:
54 "Operation timed out after 3000 milliseconds with 876 out of 876 bytes
57 It should only a timeout if the timer lapsed and we DIDN'T receive the end
58 of the response body yet.
60 - Christopher Conroy fixed a problem with RTSP and GET_PARAMETER reported
61 to us by Massimo Callegari. There's a new test case 572 that verifies this
64 - The 'ares' subtree has been removed from the source repository. It was
65 always a separate project that sort of piggybacked on the curl project since
66 the dawn of times and now the time has come for it to go stand on its own
67 legs and continue living its own life. All details on c-ares and its new
68 source code repository is found at http://c-ares.haxx.se/
70 Daniel Stenberg (23 Mar 2010)
71 - Kenny To filed the bug report #2963679 with patch to fix a problem he
72 experienced with doing multi interface HTTP POST over a proxy using
73 PROXYTUNNEL. He found a case where it would connect fine but bits.tcpconnect
74 was not set correct so libcurl didn't work properly.
76 (http://curl.haxx.se/bug/view.cgi?id=2963679)
78 - Akos Pasztory filed debian bug report #572276
79 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=572276 mentioning a problem
80 with a resource that returns chunked-encoded _and_ with a Content-Length
81 and libcurl failed to properly ignore the latter information.
83 - Hauke Duden provided an example program that made the multi interface crash.
84 His example simply used the multi interface and did first one FTP transfer
85 and after completion it used a second easy handle and did another FTP
86 transfer on the same FTP server.
88 This triggered a bug in the "delayed easy handle kill" system that curl
89 uses: when an FTP connection is left alive it must keep an easy handle
90 around internally - only for the purpose of having an easy handle when it
91 later disconnects it. The code assumed that when the easy handle was removed
92 and an internal reference was made, that version could be killed later on
93 when a new easy handle came using the same connection. This was wrong as
94 Hauke's example showed that the removed handle wasn't killed for real until
95 later. This caused a double close attempt => segfault.
97 Daniel Stenberg (22 Mar 2010)
98 - Thomas Lopatic fixed the alarm()-based DNS timeout:
100 Looking at the code of Curl_resolv_timeout() in hostip.c, I think that in
101 case of a timeout, the signal handler for SIGALRM never gets removed. I
102 think that in my case it gets executed at some point later on when execution
103 has long left Curl_resolv_timeout() or even the cURL library.
105 The code that is jumped to with siglongjmp() simply sets the error message
106 to "name lookup timed out" and then returns with CURLRESOLV_ERROR. I guess
107 that instead of simply returning without cleaning up, the code should have a
108 goto that jumps to the spot right after the call to Curl_resolv().
110 Kamil Dudka (22 Mar 2010)
111 - Douglas Steinwand contributed a patch fixing insufficient initialization in
112 Curl_clone_ssl_config()
114 Daniel Stenberg (21 Mar 2010)
115 - Ben Greear improved TFTP: the error code returning and the treatment
116 of TSIZE == 0 when uploading.
118 - We've switched from CVS to git. See http://curl.haxx.se/source.html
120 Kamil Dudka (19 Mar 2010)
121 - Improved Curl_read() to not ignore the error returned from Curl_ssl_recv().
123 Daniel Stenberg (15 Mar 2010)
124 - Constantine Sapuntzakis brought a patch:
126 The problem mentioned on Dec 10 2009
127 (http://curl.haxx.se/bug/view.cgi?id=2905220) was only partially fixed.
128 Partially because an easy handle can be associated with many connections in
129 the cache (e.g. if there is a redirect during the lifetime of the easy
130 handle). The previous patch only cleaned up the first one. The new fix now
131 removes the easy handle from all connections, not just the first one.
133 Daniel Stenberg (6 Mar 2010)
134 - Ben Greear brought a patch that fixed the rate limiting logic for TFTP when
135 the easy interface was used.
137 Daniel Stenberg (5 Mar 2010)
138 - Daniel Johnson provided fixes for building curl with the clang compiler.
140 Yang Tse (5 Mar 2010)
141 - Constantine Sapuntzakis detected and fixed a double free in builds done
142 with threaded resolver enabled (Windows default configuration) that would
143 get triggered when a curl handle is closed while doing DNS resolution.
145 Daniel Stenberg (2 Mar 2010)
146 - [Daniel Johnson] I've been trying to build libcurl with clang on Darwin and
147 ran into some issues with the GSSAPI tests in configure.ac. The tests first
148 try to determine the include dirs and libs and set CPPFLAGS and LIBS
149 accordingly. It then checks for the headers and finally sets LIBS a second
150 time, causing the libs to be included twice. The first setting of LIBS seems
151 redundant and should be left out, since the first part is otherwise just
152 about finding headers.
154 My second issue is that 'krb5-config --libs gssapi' on Darwin is less than
155 useless and returns junk that, while it happens to work with gcc, causes
156 clang to choke. For example, --libs returns $CFLAGS along with the libs,
157 which is really retarded. Simply setting 'LIBS="$LIBS -lgssapi_krb5
158 -lresolv"' on Darwin is sufficient.
160 - Based on patch provided by Jacob Moshenko, the transfer logic now properly
161 makes sure that when using sub-second timeouts, there's no final bad 1000ms
162 wait. Previously, a sub-second timeout would often make the elapsed time end
163 up the time rounded up to the nearest second (e.g. 1s for 200ms timeout)
165 - Andrei Benea filed bug report #2956698 and pointed out that the
166 CURLOPT_CERTINFO feature leaked memory due to a missing OpenSSL function
167 call. He provided the patch to fix it too.
169 http://curl.haxx.se/bug/view.cgi?id=2956698
171 - Markus Duft pointed out in bug #2961796 that even though Interix has a
172 poll() function it doesn't quite work the way we want it so we must disable
173 it, and he also provided a patch for it.
175 http://curl.haxx.se/bug/view.cgi?id=2961796
177 - Made the pingpong timeout code properly deal with the response timeout AND
178 the global timeout if set. Also, as was reported in the bug report #2956437
179 by Ryan Chan, the time stamp to use as basis for the per command timeout was
180 not set properly in the DONE phase for FTP (and not for SMTP) so I fixed
181 that just now. This was a regression compared to 7.19.7 due to the
182 conversion of FTP code over to the generic pingpong concepts.
184 http://curl.haxx.se/bug/view.cgi?id=2956437
186 Daniel Stenberg (1 Mar 2010)
187 - Ben Greear provided an update for TFTP that fixes upload.
189 - Wesley Miaw reported bug #2958179 which identified a case of looping during
190 OpenSSL based SSL handshaking even though the multi interface was used and
191 there was no good reason for it.
193 http://curl.haxx.se/bug/view.cgi?id=2958179
195 Daniel Stenberg (26 Feb 2010)
196 - Pat Ray in bug #2958474 pointed out an off-by-one case when receiving a
197 chunked-encoding trailer.
199 http://curl.haxx.se/bug/view.cgi?id=2958474
201 Daniel Fandrich (25 Feb 2010)
202 - Fixed a couple of out of memory leaks and a segfault in the SMTP & IMAP code.
204 Yang Tse (25 Feb 2010)
205 - I fixed bug report #2958074 indicating
206 (http://curl.haxx.se/bug/view.cgi?id=2958074) that curl on Windows with
207 option --trace-time did not use local time when timestamping trace lines.
208 This could also happen on other systems depending on time souurce.
210 Patrick Monnerat (22 Feb 2010)
211 - Proper handling of STARTTLS on SMTP, taking CURLUSESSL_TRY into account.
212 - SMTP falls back to RFC821 HELO when EHLO fails (and SSL is not required).
213 - Use of true local host name (i.e.: via gethostname()) when available, as
214 default argument to SMTP HELO/EHLO.
215 - Test case 804 for HELO fallback.
217 Daniel Stenberg (20 Feb 2010)
218 - Fixed the SMTP compliance by making sure RCPT TO addresses are specified
219 properly in angle brackets. Recipients provided with CURLOPT_MAIL_RCPT now
220 get angle bracket wrapping automatically by libcurl unless the recipient
221 starts with an angle bracket as then the app is assumed to deal with that
224 - I made the SMTP code expect a 250 response back from the server after the
225 full DATA has been sent, and I modified the test SMTP server to also send
226 that response. As usual, the DONE operation that is made after a completed
227 transfer is still not doable in a non-blocking way so this waiting for 250
228 is unfortunately made blockingly.
230 Yang Tse (14 Feb 2010)
231 - Overhauled test suite getpart() function. Fixing potential out of bounds
232 stack and memory overwrites triggered with huge test case definitions.
234 Daniel Stenberg (13 Feb 2010)
235 - Martin Hager reported and fixed a problem with a missing quote in libcurl.m4
237 (http://curl.haxx.se/bug/view.cgi?id=2951319)
239 - Tom Donovan fixed the CURL_FORMAT_* defines when building with cmake.
241 (http://curl.haxx.se/bug/view.cgi?id=2951269)
243 Daniel Stenberg (12 Feb 2010)
244 - Jack Zhang reported a problem with SMTP: we wrongly used multiple addresses
245 in the same RCPT TO line, when they should be sent in separate single
246 commands. I updated test case 802 to verify this.
248 - I also fixed a bad use of my_setopt_str() of CURLOPT_MAIL_RCPT in the curl
249 tool which made it try to output it as string for the --libcurl feature
250 which could lead to crashes.
252 Yang Tse (11 Feb 2010)
253 - Steven M. Schweda fixed VMS builder bad behavior when used in a batch job,
254 removed obsolete batch_compile.com and defines.com and updated VMS readme.
256 Version 7.20.0 (9 February 2010)
258 Daniel Stenberg (9 Feb 2010)
259 - When downloading compressed content over HTTP and the app asked libcurl to
260 automatically uncompress it with the CURLOPT_ENCODING option, libcurl could
261 wrongly provide the callback with more data than the maximum documented
262 amount. An application could thus get tricked into badness if the maximum
263 limit was trusted to be enforced by libcurl itself (as it is documented).
265 This is further detailed and explained in the libcurl security advisory
268 http://curl.haxx.se/docs/adv_20100209.html
270 Daniel Fandrich (3 Feb 2010)
271 - Changed the Watcom makefiles to make them easier to keep in sync with
272 Makefile.inc since that can't be included directly.
274 Yang Tse (2 Feb 2010)
275 - Symbol CURL_FORMAT_OFF_T now obsoleted, will be removed in a future release,
276 symbol will not be available when building with CURL_NO_OLDIES defined. Use
277 of CURL_FORMAT_CURL_OFF_T is preferred since 7.19.0
279 Daniel Stenberg (1 Feb 2010)
280 - Using the multi_socket API, it turns out at times it seemed to "forget"
281 connections (which caused a hang). It turned out to be an existing (7.19.7)
282 bug in libcurl (that's been around for a long time) and it happened like
285 The app calls curl_multi_add_handle() to add a new easy handle, libcurl will
286 then set it to timeout in 1 millisecond so libcurl will tell the app about
289 The app's timeout fires off that there's a timeout, the app calls libcurl as
290 we so often document it:
293 res = curl_multi_socket_action(... TIMEOUT ...);
294 } while(CURLM_CALL_MULTI_PERFORM == res);
296 And this is the problem number one:
298 When curl_multi_socket_action() is called with no specific handle, but only
299 a timeout-action, it will *only* perform actions within libcurl that are
300 marked to run at this time. In this case, the request would go from INIT to
301 CONNECT and return CURLM_CALL_MULTI_PERFORM. When the app then calls libcurl
302 again, there's no timer set for this handle so it remains in the CONNECT
303 state. The CONNECT state is a transitional state in libcurl so it reports no
304 sockets there, and thus libcurl never tells the app anything more about that
305 easy handle/connection.
307 libcurl _does_ set a 1ms timeout for the handle at the end of
308 multi_runsingle() if it returns CURLM_CALL_MULTI_PERFORM, but since the loop
309 is instant the new job is not ready to run at that point (and there's no
310 code that makes libcurl call the app to update the timout for this new
311 timeout). It will simply rely on that some other timeout will trigger later
312 on or that something else will update the timeout callback. This makes the
313 bug fairly hard to repeat.
315 The fix made to adress this issue:
317 We introduce a loop in lib/multi.c around all calls to multi_runsingle() and
318 simply check for CURLM_CALL_MULTI_PERFORM internally. This has the added
319 benefit that this goes in line with my long-term wishes to get rid of the
320 CURLM_CALL_MULTI_PERFORM all together from the public API.
322 The downside of this fix, is that the counter we return in 'running_handles'
323 in several of our public functions then gets a slightly new and possibly
324 confusing behavior during times:
326 If an app adds a handle that fails to connect (very quickly) it may just
327 as well never appear as a 'running_handle' with this fix. Previously it
328 would first bump the counter only to get it decreased again at next call.
329 Even I have used that change in handle counter to signal "end of a
330 transfer". The only *good* way to find the end of a individual transfer
331 is calling curl_multi_info_read() to see if it returns one.
333 Of course, if the app previously did the looping before it checked the
334 counter, it really shouldn't be any new effect.
336 Yang Tse (26 Jan 2010)
337 - Constantine Sapuntzakis' and Joshua Kwan's work done in the last four months
338 relative to the asynchronous DNS lookups, along with with some integration
339 adjustments I have done are finally committed to CVS.
341 Currently these enhancements will benefit builds done using c-ares on any
342 platform as well as Windows builds using the default threaded resolver.
344 This release does not make generally available POSIX threaded DNS lookups
345 yet. There is no configure option to enable this feature yet. It is possible
346 to experimantally try this feature running configure with compiler flags that
347 make simultaneous definition of preprocessor symbols USE_THREADS_POSIX and
348 HAVE_PTHREAD_H, as well as whatever reentrancy compiler flags and linker ones
349 are required to link and properly use pthread_* functions on each platform.
351 Daniel Stenberg (26 Jan 2010)
352 - Mike Crowe made libcurl return CURLE_COULDNT_RESOLVE_PROXY when it is the
353 proxy that cannot be resolved when using c-ares. This matches the behaviour
354 when not using c-ares.
356 Björn Stenberg (23 Jan 2010)
357 - Added a new flag: -J/--remote-header-name. This option tells the
358 -O/--remote-name option to use the server-specified Content-Disposition
359 filename instead of extracting a filename from the URL.
361 Daniel Stenberg (21 Jan 2010)
362 - Chris Conroy brought support for RTSP transfers, and with it comes 8(!) new
363 libcurl options for controlling what to get and how to receive posssibly
364 interleaved RTP data.
366 Daniel Stenberg (20 Jan 2010)
367 - As was pointed out on the http-state mailing list, the order of cookies in a
368 HTTP Cookie: header _needs_ to be sorted on the path length in the cases
369 where two cookies using the same name are set more than once using
370 (overlapping) paths. Realizing this, identically named cookies must be
371 sorted correctly. But detecting only identically named cookies and take care
372 of them individually is harder than just to blindly and unconditionally sort
373 all cookies based on their path lengths. All major browsers also already do
374 this, so this makes our behavior one step closer to them in the cookie area.
376 Test case 8 was the only one that broke due to this change and I updated it
379 Daniel Stenberg (19 Jan 2010)
380 - David McCreedy brought a fix and a new test case (129) to make libcurl work
381 again when downloading files over FTP using ASCII and it turns out that the
382 final size of the file is not the same as the initial size the server
383 reported. This is very common since servers don't take the newline
384 conversions into account.
386 Kamil Dudka (14 Jan 2010)
387 - Suppressed side effect of OpenSSL configure checks, which prevented NSS from
388 being properly detected under certain circumstances. It had been caused by
389 strange behavior of pkg-config when handling PKG_CONFIG_LIBDIR. pkg-config
390 distinguishes among empty and non-existent environment variable in that case.
392 Daniel Stenberg (12 Jan 2010)
393 - Gil Weber reported a peculiar flaw with the multi interface when doing SFTP
394 transfers: curl_multi_fdset() would return -1 and not set and file
395 descriptors several times during a transfer of a single file. It turned out
396 to be due to two different flaws now fixed. Gil's excellent recipe helped me
399 Daniel Stenberg (11 Jan 2010)
400 - Made sure that the progress callback is repeatedly called at a regular
401 interval even during very slow connects.
403 - The tests/runtests.pl script now checks to see if the test case that runs is
404 present in the tests/data/Makefile.am and outputs a notice message on the
405 screen if not. Each test file has to be included in that Makefile.am to get
406 included in release archives and forgetting to add files there is a common
407 mistake. This is an attempt to make it harder to forget.
409 Daniel Stenberg (9 Jan 2010)
410 - Johan van Selst found and fixed a OpenSSL session ref count leak:
412 ossl_connect_step3() increments an SSL session handle reference counter on
413 each call. When sessions are re-used this reference counter may be
414 incremented many times, but it will be decremented only once when done (by
415 Curl_ossl_session_free()); and the internal OpenSSL data will not be freed
416 if this reference count remains positive. When a session is re-used the
417 reference counter should be corrected by explicitly calling
418 SSL_SESSION_free() after each consecutive SSL_get1_session() to avoid
419 introducing a memory leak.
421 (http://curl.haxx.se/bug/view.cgi?id=2926284)
423 Daniel Stenberg (7 Jan 2010)
424 - Make sure the progress callback is called repeatedly even during very slow
425 name resolves when c-ares is used for resolving.
427 Claes Jakobsson (6 Jan 2010)
428 - Julien Chaffraix fixed so that the fragment part in an URL is not sent
429 to the server anymore.
431 Kamil Dudka (3 Jan 2010)
432 - Julien Chaffraix eliminated a duplicated initialization in singlesocket().
434 Daniel Stenberg (2 Jan 2010)
435 - Make curl support --ssl and --ssl-reqd instead of the previous FTP-specific
436 versions --ftp-ssl and --ftp-ssl-reqd as these options are now used to
437 control SSL/TLS for IMAP, POP3 and SMTP as well in addition to FTP. The old
438 option names are still working but the new ones are the ones listed and
441 Daniel Stenberg (1 Jan 2010)
442 - Ingmar Runge enhanced libcurl's FTP engine to support the PRET command. This
443 command is a special "hack" used by the drftpd server, but even though it is
444 a custom extension I've deemed it fine to add to libcurl since this server
445 seems to survive and people keep using it and want libcurl to support
446 it. The new libcurl option is named CURLOPT_FTP_USE_PRET, and it is also
447 usable from the curl tool with --ftp-pret. Using this option on a server
448 that doesn't support this command will make libcurl fail.
450 I added test cases 1107 and 1108 to verify the functionality.
452 The PRET command is documented at
453 http://www.drftpd.org/index.php/Distributed_PASV
455 Yang Tse (30 Dec 2009)
456 - Steven M. Schweda improved VMS build system, and Craig A. Berry helped
457 with the patch and testing.
459 Daniel Stenberg (26 Dec 2009)
460 - Renato Botelho and Peter Pentchev brought a patch that makes the libcurl
461 headers work correctly even on FreeBSD systems before v8.
463 (http://curl.haxx.se/bug/view.cgi?id=2916915)
465 Daniel Stenberg (17 Dec 2009)
466 - David Byron fixed Curl_ossl_cleanup to actually call ENGINE_cleanup when
469 - Follow-up fix for the proxy fix I did for Jon Nelson's bug. It turned out I
470 was a bit too quick and broke test case 1101 with that change. The order of
471 some of the setups is sensitive. I now changed it slightly again to make
472 sure we do them in this order:
474 1 - parse URL and figure out what protocol is used in the URL
475 2 - prepend protocol:// to URL if missing
476 3 - parse name+password off URL, which needs to know what protocol is used
477 (since only some allows for name+password in the URL)
478 4 - figure out if a proxy should be used set by an option
479 5 - if no proxy option, check proxy environment variables
480 6 - run the protocol-specific setup function, which needs to have the proxy
483 Daniel Stenberg (15 Dec 2009)
484 - Jon Nelson found a regression that turned out to be a flaw in how libcurl
485 detects and uses proxies based on the environment variables. If the proxy
486 was given as an explicit option it worked, but due to the setup order
487 mistake proxies would not be used fine for a few protocols when picked up
488 from '[protocol]_proxy'. Obviously this broke after 7.19.4. I now also added
489 test case 1106 that verifies this functionality.
491 (http://curl.haxx.se/bug/view.cgi?id=2913886)
493 Daniel Stenberg (12 Dec 2009)
494 - IMAP, POP3 and SMTP support and their TLS versions (including IMAPS, POP3S
495 and SMTPS) are now supported. The current state may not yet be solid, but
496 the foundation is in place and the test suite has some initial support for
497 these protocols. Work will now persue to make them nice libcurl citizens
500 The work with supporting these new protocols was sponsored by
501 networking4all.com - thanks!
503 Daniel Stenberg (10 Dec 2009)
504 - Siegfried Gyuricsko found out that the curl manual said --retry would retry
505 on FTP errors in the transient 5xx range. Transient FTP errors are in the
506 4xx range. The code itself only tried on 5xx errors that occured _at login_.
507 Now the retry code retries on all FTP transfer failures that ended with a
510 (http://curl.haxx.se/bug/view.cgi?id=2911279)
512 - Constantine Sapuntzakis figured out a case which would lead to libcurl
513 accessing alredy freed memory and thus crash when using HTTPS (with
514 OpenSSL), multi interface and the CURLOPT_DEBUGFUNCTION and a certain order
515 of cleaning things up. I fixed it.
517 (http://curl.haxx.se/bug/view.cgi?id=2905220)
519 Daniel Stenberg (7 Dec 2009)
520 - Martin Storsjo made libcurl use the Expect: 100-continue header for posts
521 with unknown size. Previously it was only used for posts with a known size
522 larger than 1024 bytes.
524 Daniel Stenberg (1 Dec 2009)
525 - If the Expect: 100-continue header has been set by the application through
526 curl_easy_setopt with CURLOPT_HTTPHEADER, the library should set
527 data->state.expect100header accordingly - the current code (in 7.19.7 at
528 least) doesn't handle this properly. Martin Storsjo provided the fix!
530 Yang Tse (28 Nov 2009)
531 - Added Diffie-Hellman parameters to several test harness certificate files in
532 PEM format. Required by several stunnel versions used by our test harness.
534 Daniel Stenberg (28 Nov 2009)
535 - Markus Koetter provided a polished and updated version of Chad Monroe's TFTP
536 rework patch that now integrates TFTP properly into libcurl so that it can
537 be used non-blocking with the multi interface and more. BLKSIZE also works.
539 The --tftp-blksize option was added to allow setting the TFTP BLKSIZE from
542 Daniel Stenberg (26 Nov 2009)
543 - Extended and fixed the change I did on Dec 11 for the the progress
544 meter/callback during FTP command/response sequences. It turned out it was
545 really lame before and now the progress meter SHOULD get called at least
548 Daniel Stenberg (23 Nov 2009)
549 - Bjorn Augustsson reported a bug which made curl not report any problems even
550 though it failed to write a very small download to disk (done in a single
551 fwrite call). It turned out to be because fwrite() returned success, but
552 there was insufficient error-checking for the fclose() call which tricked
553 curl to believe things were fine.
555 Yang Tse (23 Nov 2009)
556 - David Byron modified Makefile.dist vc8 and vc9 targets in order to allow
557 finer granularity control when generating src and lib makefiles.
559 Yang Tse (22 Nov 2009)
560 - I modified configure to force removal of the curlbuild.h file included in
561 distribution tarballs for use by non-configure systems. As intended, this
562 would get overwriten when doing in-tree builds. But VPATH builds would end
563 having two curlbuild.h files, one in the source tree and another in the
564 build tree. With the modification I introduced 5 Nov 2009 this could become
565 an issue when running libcurl's test suite.
567 Daniel Stenberg (20 Nov 2009)
568 - Constantine Sapuntzakis identified a write after close, as the sockets were
569 closed by libcurl before the SSL lib were shutdown and they may write to its
570 socket. Detected to at least happen with OpenSSL builds.
572 - Jad Chamcham pointed out a bug with connection re-use. If a connection had
573 CURLOPT_HTTPPROXYTUNNEL enabled over a proxy, a subsequent request using the
574 same proxy with the tunnel option disabled would still wrongly re-use that
575 previous connection and the outcome would only be badness.
577 Yang Tse (18 Nov 2009)
578 - I modified the memory tracking system to make it intolerant with zero sized
579 malloc(), calloc() and realloc() function calls.
581 Daniel Stenberg (17 Nov 2009)
582 - Constantine Sapuntzakis provided another fix for the DNS cache that could
583 end up with entries that wouldn't time-out:
585 1. Set up a first web server that redirects (307) to a http://server:port
587 2. Have curl connect to the first web server using curl multi
589 After the curl_easy_cleanup call, there will be curl dns entries hanging
590 around with in_use != 0.
592 (http://curl.haxx.se/bug/view.cgi?id=2891591)
594 - Marc Kleine-Budde fixed: curl saved the LDFLAGS set during configure into
595 its pkg-config file. So -Wl stuff ended up in the .pc file, which is really
596 bad, and breaks if there are multiple -Wl in our LDFLAGS (which are in
597 PTXdist). bug #2893592 (http://curl.haxx.se/bug/view.cgi?id=2893592)
599 Kamil Dudka (15 Nov 2009)
600 - David Byron improved the configure script to use pkg-config to find OpenSSL
601 (and in particular the list of required libraries) even if a path is given
602 as argument to --with-ssl
604 Yang Tse (15 Nov 2009)
605 - I removed enable-thread / disable-thread configure option. These were only
606 placebo options. The library is always built as thread safe as possible on
609 Claes Jakobsson (14 Nov 2009)
610 - curl-config now accepts '--configure' to see what arguments was
611 passed to the configure script when building curl.
613 Daniel Stenberg (14 Nov 2009)
614 - Claes Jakobsson restored the configure functionality to detect NSS when
615 --with-nss is set but not "yes".
617 I think we can still improve that to check for pkg-config in that path etc,
618 but at least this patch brings back the same functionality we had before.
620 - Camille Moncelier added support for the file type SSL_FILETYPE_ENGINE for
621 the client certificate. It also disable the key name test as some engines
622 can select a private key/cert automatically (When there is only one key
623 and/or certificate on the hardware device used by the engine)
625 Yang Tse (14 Nov 2009)
626 - Constantine Sapuntzakis provided the fix that ensures that an SSL connection
627 won't be reused unless protection level for peer and host verification match.
629 I refactored how preprocessor symbol _THREAD_SAFE definition is done.
631 Kamil Dudka (12 Nov 2009)
632 - Kevin Baughman provided a fix preventing libcurl-NSS from crash on doubly
633 closed NSPR descriptor. The issue was hard to find, reported several times
634 before and always closed unresolved. More info at the RH bug:
635 https://bugzilla.redhat.com/534176
637 - libcurl-NSS now tries to reconnect with TLS disabled in case it detects
638 a broken TLS server. However it does not happen if SSL version is selected
639 manually. The approach was originally taken from PSM. Kaspar Brand helped me
640 to complete the patch. Original bug reports:
641 https://bugzilla.redhat.com/525496
642 https://bugzilla.redhat.com/527771
644 Yang Tse (12 Nov 2009)
645 - I modified configure script to make the getaddrinfo function check also
646 verify if the function is thread safe.
648 Yang Tse (11 Nov 2009)
649 - Marco Maggi reported that compilation failed when configured --with-gssapi
650 and GNU GSS installed due to a missing mutual exclusion of header files in
651 the Kerberos 5 code path. He also verified that my patch worked for him.
653 Daniel Stenberg (11 Nov 2009)
654 - Constantine Sapuntzakis posted bug #2891595
655 (http://curl.haxx.se/bug/view.cgi?id=2891595) which identified how an entry
656 in the DNS cache would linger too long if the request that added it was in
657 use that long. He also provided the patch that now makes libcurl capable of
658 still doing a request while the DNS hash entry may get timed out.
660 - Christian Schmitz noticed that the progress meter/callback was not properly
661 used during the FTP connection phase (after the actual TCP connect), while
662 it of course should be. I also made the speed check get called correctly so
663 that really slow servers will trigger that properly too.
665 Kamil Dudka (5 Nov 2009)
666 - Dropped misleading timeouts in libcurl-NSS and made sure the SSL socket works
667 in non-blocking mode.
669 Yang Tse (5 Nov 2009)
670 - I removed leading 'curl' path on the 'curlbuild.h' include statement in
671 curl.h, adjusting auto-makefiles include path, to enhance portability to
672 OS's without an orthogonal directory tree structure such as OS/400.
674 Daniel Stenberg (4 Nov 2009)
675 - I fixed several problems with the transfer progress meter. It showed the
676 wrong percentage for small files, most notable for <1000 bytes and could
677 easily end up showing more than 100% at the end. It also didn't show any
678 percentage, transfer size or estimated transfer times when transferring
681 Version 7.19.7 (4 November 2009)
683 Daniel Stenberg (2 Nov 2009)
684 - As reported independent by both Stan van de Burgt and Didier Brisebourg,
685 CURLINFO_SIZE_DOWNLOAD (the -w variable size_download) didn't work when
686 getting data from ldap!
688 Daniel Stenberg (31 Oct 2009)
689 - Gabriel Kuri reported a problem with CURLINFO_CONTENT_LENGTH_DOWNLOAD if the
690 download was 0 bytes, as libcurl would then return the size as unknown (-1)
691 and not 0. I wrote a fix and test case 566 to verify it.
693 Daniel Stenberg (30 Oct 2009)
694 - Liza Alenchery mentioned a problem with re-used SCP connection when a bad
695 auth is used, as it caused a crash. I failed to repeat the issue, but still
696 made a change that now forces the TCP connection used for a freed SCP
697 session to get closed and not be re-used.
699 - "Tom" posted a bug report that mentioned how libcurl did wrong when doing a
700 POST using a read callback, with Digest authentication and
701 "Transfer-Encoding: chunked" enforced. I would then cause the first request
702 to be wrongly sent and then basically hang until the server closed the
703 connection. I fixed the problem and added test case 565 to verify it.
705 Daniel Stenberg (25 Oct 2009)
706 - Dima Barsky made the curl cookie parser accept cookies even with blank or
707 unparsable expiry dates and then treat them as session cookies - previously
708 libcurl would reject cookies with a date format it couldn't parse. Research
709 shows that the major browser treat such cookies as session cookies. I
710 modified test 8 and 31 to verify this.
712 Daniel Stenberg (21 Oct 2009)
713 - Attempt to use pkg-config for finding out libssh2 installation details
716 - A patch in bug report #2883177 (http://curl.haxx.se/bug/view.cgi?id=2883177)
717 by Johan van Selst introduced the --crlfile option to curl, which makes curl
718 tell libcurl about a file with CRL (certificate revocation list) data to
721 Daniel Stenberg (18 Oct 2009)
722 - Ray Dassen provided a patch in Debian's bug tracker (bug number #551461)
723 that now makes curl_getdate(3) actually handles RFC 822 formatted dates that
724 use the "single letter military timezones".
725 http://www.rfc-ref.org/RFC-TEXTS/822/chapter5.html has the details.
727 - Fixed memory leak in the SCP/SFTP code as it never freed the knownhosts
730 - John Dennis filed bug report #2873666
731 (http://curl.haxx.se/bug/view.cgi?id=2873666) which identified a problem
732 which made libcurl loop infinitely when given incorrect credentials when
733 using HTTP GSS negotiate authentication. He also provided a small and simple
736 - Kevin Baughman found a double close() problem with libcurl-NSS, as when
737 libcurl called NSS to close the SSL "session" it also closed the actual
740 Yang Tse (17 Oct 2009)
741 - Bug report #2866724 indicated
742 (http://curl.haxx.se/bug/view.cgi?id=2866724) that curl on Windows failed
743 when writing files whose file names originally contained characters which
744 are not valid for file names on Windows. Dan Fandrich provided an initial
745 patch and another revised one to fix this issue.
747 Daniel Stenberg (1 Oct 2009)
748 - Tom Mueller correctly reported in bug report #2870221
749 (http://curl.haxx.se/bug/view.cgi?id=2870221) that libcurl returned an
750 incorrect return code from the internal trynextip() function which caused
751 him grief. This is a regression that was introduced in 7.19.1 and I find it
752 strange it hasn't hit us harder, but I won't persue into figuring out
755 - Constantine Sapuntzakis: The current implementation will always set
756 SO_SNDBUF to CURL_WRITE_SIZE even if the SO_SNDBUF starts out larger. The
757 patch doesn't do a setsockopt if SO_SNDBUF is already greater than
758 CURL_WRITE_SIZE. This should help folks who have set up their computer with
761 Daniel Stenberg (27 Sep 2009)
762 - I introduced a maximum limit for received HTTP headers. It is controlled by
763 the define CURL_MAX_HTTP_HEADER which is even exposed in the public header
764 file to allow for users to fairly easy rebuild libcurl with a modified
765 limit. The rationale for a fixed limit is that libcurl is realloc()ing a
766 buffer to be able to put a full header into it, so that it can call the
767 header callback with the entire header, but that also risk getting it into
768 trouble if a server by mistake or willingly sends a header that is more or
769 less without an end. The limit is set to 100K.
771 Daniel Stenberg (26 Sep 2009)
772 - John P. McCaskey posted a bug report that showed how libcurl did wrong when
773 saving received cookies with no given path, if the path in the request had a
774 query part. That is means a question mark (?) and characters on the right
775 side of that. I wrote test case 1105 and fixed this problem.
777 Kamil Dudka (26 Sep 2009)
778 - Implemented a protocol independent way to specify blocking direction, used by
779 transfer.c for blocking. It is currently used only by SCP and SFTP protocols.
780 This enhancement resolves an issue with 100% CPU usage during SFTP upload,
783 Daniel Stenberg (25 Sep 2009)
784 - Chris Mumford filed bug report #2861587
785 (http://curl.haxx.se/bug/view.cgi?id=2861587) identifying that libcurl used
786 the OpenSSL function X509_load_crl_file() wrongly and failed if it would
787 load a CRL file with more than one certificate within. This is now fixed.
789 Daniel Stenberg (16 Sep 2009)
790 - Sven Anders reported that we introduced a cert verfication flaw for OpenSSL-
791 powered libcurl in 7.19.6. If there was a X509v3 Subject Alternative Name
792 field in the certficate it had to match and so even if non-DNS and non-IP
793 entry was present it caused the verification to fail.
795 Daniel Fandrich (15 Sep 2009)
796 - Moved the libssh2 checks after the SSL library checks. This helps when
797 statically linking since libssh2 needs the SSL library link flags to be
798 set up already to satisfy its dependencies. This wouldn't be necessary if
799 the libssh2 configure check was changed to use pkg-config since the
800 --static flag would add the dependencies automatically.
802 Yang Tse (14 Sep 2009)
803 - Revert Joshua Kwan's patch committed 11 Sep 2009.
805 Some systems poll function sets POLLHUP in revents without setting
806 POLLIN, and sets POLLERR without setting POLLIN and POLLOUT. In some
807 libcurl code execution paths this could trigger busy wait loops with
808 high CPU usage until a timeout condition aborted the loop.
810 The reverted patch addressed the above issue for a very specific case,
811 when awaiting c-ares to resolve. A libcurl-wide fix for Curl_poll now
814 Guenter Knauf (11 Sep 2009)
815 - Joshua Kwan provided a patch to pass POLLERR / POLLHUP back to c-ares.
816 This fixes a loop problem with high CPU usage.
818 Daniel Stenberg (10 Sep 2009)
819 - Claes Jakobsson fixed a problem with cookie expiry dates at exctly the epoch
820 start second "Thu Jan 1 00:00:00 GMT 1970" as the date parser then returns 0
821 which internally then is treated as a session cookie. That particular date
822 is now made to get the value of 1.
824 Daniel Stenberg (2 Sep 2009)
825 - Daniel Johnson found a flaw in the code converting sftp-errors to libcurl
828 Daniel Stenberg (1 Sep 2009)
829 - Peter Sylvester made a debug feature for Curl_resolv() that now will force
830 libcurl to resolve 'localhost' whatever name you use in the URL *if* you set
831 the --interface option to (exactly) "LocalHost". This will enable us to
832 write tests for custom hosts names but still use a local host server.
834 - configure now tries to use pkg-config for a number of sub-dependencies even
835 when cross-compiling. The key to success is then you properly setup
836 PKG_CONFIG_PATH before invoking configure.
838 I also improved how NSS is detected by trying nss-config if pkg-config isn't
839 present, and as a last resort just use the lib name and force the user to
840 setup the LIBS/LDFLAGS/CFLAGS etc properly. The previous last resort would
841 add a range of various libs that would almost never be quite correct.
843 Daniel Stenberg (31 Aug 2009)
844 - When using the multi interface with FTP and you asked for NOBODY, you did no
845 QUOTE commands and the request used the same path as the connection had
846 already changed to, it would decide that no commands would be necessary for
847 the "DO" action and that was not handled properly but libcurl would instead
850 Kamil Dudka (28 Aug 2009)
851 - Improved error message for not matching certificate subject name in
852 libcurl-NSS. Originally reported at:
853 https://bugzilla.redhat.com/show_bug.cgi?id=516056#c9
855 Patrick Monnerat (24 Aug 2009)
856 - Introduced a SYST-based test to properly set-up name format when dealing
857 with the OS/400 FTP server.
859 - Fixed an ftp_readresp() bug preventing detection of failing control socket
860 and causing FTP client to loop forever.
862 Daniel Stenberg (24 Aug 2009)
863 - Marc de Bruin pointed out that configure --with-gnutls=PATH didn't work
864 properly and provided a fix. http://curl.haxx.se/bug/view.cgi?id=2843008
866 - Eric Wong introduced support for the new option -T. (dot) that makes curl
867 read stdin in a non-blocking fashion. This also brings back -T- (minus) to
868 the previous blocking behavior since it could break stuff for people at
871 Michal Marek (21 Aug 2009)
872 - With CURLOPT_PROXY_TRANSFER_MODE, avoid sending invalid URLs like
873 ftp://example.com;type=i if the user specified ftp://example.com without the
876 Daniel Stenberg (21 Aug 2009)
877 - Andre Guibert de Bruet pointed out a missing return code check for a
878 strdup() that could lead to segfault if it returned NULL. I extended his
879 suggest patch to now have Curl_retry_request() return a regular return code
880 and better check that.
882 - Lots of good work by Krister Johansen, mostly related to pipelining:
884 Fix SIGSEGV on free'd easy_conn when pipe unexpectedly breaks
885 Fix data corruption issue with re-connected transfers
886 Fix use after free if we're completed but easy_conn not NULL
888 Kamil Dudka (13 Aug 2009)
889 - Changed NSS code to not ignore the value of ssl.verifyhost and produce more
890 verbose error messages. Originally reported at:
891 https://bugzilla.redhat.com/show_bug.cgi?id=516056
893 Daniel Stenberg (12 Aug 2009)
894 - Karl Moerder fixed the Makefile.vc* makefiles to include the new file
895 nonblock.c so that they work fine again
897 - I expanded test 517 with a bunch of more dates that originate from the
898 Chrome browser test suite. It turns out most of them get parsed the same
901 Version 7.19.6 (12 August 2009)
903 Daniel Stenberg (12 Aug 2009)
904 - Carsten Lange reported a bug and provided a patch for TFTP upload and the
905 sending of the TSIZE option. I don't like fixing bugs just hours before
906 a release, but since it was broken and the patch fixes this for him I decided
909 Daniel Stenberg (11 Aug 2009)
910 - Peter Sylvester made the HTTPS test server use specific certificates for
911 each test, so that the test suite can now be used to actually test the
912 verification of cert names etc. This made an error show up in the OpenSSL-
913 specific code where it would attempt to match the CN field even if a
914 subjectAltName exists that doesn't match. This is now fixed and verified
917 - Benbuck Nason posted the bug report #2835196
918 (http://curl.haxx.se/bug/view.cgi?id=2835196), fixing a few compiler
919 warnings when mixing ints and bools.
921 Daniel Fandrich (10 Aug 2009)
922 - Fixed a memory leak in the FTP code and an off-by-one heap buffer overflow.
924 Daniel Fandrich (9 Aug 2009)
925 - Fixed some memory leaks in the command-line tool that caused most of the
926 torture tests to fail.
928 Daniel Stenberg (2 Aug 2009)
929 - Curt Bogmine reported a problem with SNI enabled on a particular server. We
930 should introduce an option to disable SNI, but as we're in feature freeze
931 now I've addressed the obvious bug here (pointed out by Peter Sylvester): we
932 shouldn't try to enable SNI when SSLv2 or SSLv3 is explicitly selected.
933 Code for OpenSSL and GnuTLS was fixed. NSS doesn't seem to have a particular
934 option for SNI, or are we simply not using it?
936 Daniel Stenberg (1 Aug 2009)
937 - Scott Cantor posted the bug report #2829955
938 (http://curl.haxx.se/bug/view.cgi?id=2829955) mentioning the recent SSL cert
939 verification flaw found and exploited by Moxie Marlinspike. The presentation
940 he did at Black Hat is available here:
941 https://www.blackhat.com/html/bh-usa-09/bh-usa-09-archives.html#Marlinspike
943 Apparently at least one CA allowed a subjectAltName or CN that contain a
944 zero byte, and thus clients that assumed they would never have zero bytes
945 were exploited to OK a certificate that didn't actually match the site. Like
946 if the name in the cert was "example.com\0theatualsite.com", libcurl would
947 happily verify that cert for example.com.
949 libcurl now better uses the length of the extracted name, not using the zero
950 termination for getting the string length.
952 This fixing only made and needed in OpenSSL interfacing code.
954 - Tanguy Fautre pointed out that OpenSSL's function RAND_screen() (present
955 only in some OpenSSL installs - like on Windows) isn't thread-safe and we
956 agreed that moving it to the global_init() function is a decent way to deal
959 - Alexander Beedie provided the patch for a noproxy problem: If I have set
960 CURLOPT_NOPROXY to "*", or to a host that should not use a proxy, I actually
961 could still end up using a proxy if a proxy environment variable was set.
963 Daniel Stenberg (27 Jul 2009)
964 - All the quote options (CURLOPT_QUOTE, CURLOPT_POSTQUOTE and
965 CURLOPT_PREQUOTE) now accept a preceeding asterisk before the command to
966 send when using FTP, as a sign that libcurl shall simply ignore the response
967 from the server instead of treating it as an error. Not treating a 400+ FTP
968 response code as an error means that failed commands will not abort the
969 chain of commands, nor will they cause the connection to get disconnected.
971 Daniel Stenberg (26 Jul 2009)
972 - Johan van Selst posted bug report #2825989
973 (http://curl.haxx.se/bug/view.cgi?id=2825989) pointing out that
974 OpenSSL-powered libcurl didn't support the SHA-2 digest algorithm, and
975 provided the solution too: to use OpenSSL_add_all_algorithms() in addition
976 to the older SSLeay_* alternative. OpenSSL_add_all_algorithms was added in
979 Daniel Stenberg (23 Jul 2009)
980 - Added CURLOPT_SSH_KNOWNHOSTS, CURLOPT_SSH_KEYFUNCTION, CURLOPT_SSH_KEYDATA.
981 They introduce known_host support for SSH keys to libcurl. See docs for
982 details. Note that this feature depends on a new enough libssh2 version, to
983 be supported in libssh2 1.2 and later (or current git repo at this time).
985 Michal Marek (22 Jul 2009)
986 - David Binderman found a memory and fd leak in lib/gtls.c:load_file()
987 (https://bugzilla.novell.com/523919). When looking at the code, I found that
988 also the ptr pointer can leak.
990 Kamil Dudka (20 Jul 2009)
991 - Claes Jakobsson improved the support for client certificates handling in
992 NSS-powered libcurl. Now the client certificates can be selected
993 automatically by a NSS built-in hook. Additionally pre-login to all PKCS11
994 slots is no more performed. It used to cause problems with HW tokens.
996 - Fixed reference counting for NSS client certificates. Now the PEM reader
997 module should be always properly unloaded on Curl_nss_cleanup(). If the
998 unload fails though, libcurl will try to reuse the already loaded instance.
1000 Daniel Fandrich (15 Jul 2009)
1001 - Added nonblock.c to the non-automake makefiles (note that the dependencies
1002 in the Watcom makefiles aren't quite correct).
1004 Michal Marek (15 Jul 2009)
1005 - Changed the description of CURLINFO_OS_ERRNO to make it clear that the
1006 errno is not reset on success.
1008 Guenter Knauf (14 Jul 2009)
1009 - renamed generated config.h to curl_config.h to avoid any future clashes
1010 with config.h from other projects.
1012 Daniel Stenberg (9 Jul 2009)
1013 - Eric Wong introduced curlx_nonblock() that the curl tool now (re-)uses for
1014 setting a file descriptor non-blocking. Used by the functionality Eric
1015 himself brough on June 15th.
1017 Daniel Stenberg (8 Jul 2009)
1018 - Constantine Sapuntzakis posted bug report #2813123
1019 (http://curl.haxx.se/bug/view.cgi?id=2813123) and an a patch that fixes the
1022 Url A is accessed using auth. Url A redirects to Url B (on a different
1023 server0. Url B reuses a persistent connection. Url B has auth, even though
1024 it's on a different server.
1026 Note: if Url B does not reuse a persistent connection, auth is not sent.
1030 data->state.first_host is not initialized becuase Curl_http_connect is not
1031 called when a connection is reused.
1035 move initialization of data->state.first_host to Curl_http. No code before
1036 Curl_http uses data->state.first_host anyway.
1038 Guenter Knauf (4 Jul 2009)
1039 - Markus Koetter provided a patch to avoid getnameinfo() usage which broke a
1040 couple of both IPv4 and IPv6 autobuilds.
1042 Daniel Stenberg (29 Jun 2009)
1043 - Markus Koetter made CURLOPT_FTPPORT (and curl's -P/--ftpport) support a port
1044 range if given colon-separated after the host name/address part. Like
1045 "192.168.0.1:2000-10000"
1047 - Modified the separators used for CURLOPT_CERTINFO in multi-part outputs. I
1048 don't know how they got wrong in the first place, but using this output
1049 format makes it possible to quite easily separate the string into an array
1052 Daniel Fandrich (16 June 2009)
1053 - Added a few more compiler warning options for gcc.
1055 Daniel Stenberg (16 Jun 2009)
1056 - Reuven Wachtfogel made curl -o - properly produce a binary output on windows
1057 (no newline translations). Use -B/--use-ascii if you rather get the ascii
1060 Michal Marek (16 Jun 2009)
1061 - When doing non-anonymous ftp via http proxies and the password is not
1062 provided in the url, add it there (squid needs this).
1064 Daniel Stenberg (15 Jun 2009)
1065 - Eric Wong's patch:
1067 This allows curl(1) to be used as a client-side tunnel for arbitrary stream
1068 protocols by abusing chunked transfer encoding in both the HTTP request and
1069 HTTP response. This requires server support for sending a response while a
1070 request is still being read, of course.
1072 If attempting to read from stdin returns EAGAIN, then we pause our sender.
1073 This leaves curl to attempt to read from the socket while reading from stdin
1074 (and thus sending) is paused.
1076 This change was needed to allow successfully tunneling the git protocol over
1077 HTTP (--no-buffer is needed, as well).
1079 Patrick Monnerat (15 Jun 2009)
1080 - Replaced use of standard C library rand()/srand() by our own pseudo-random
1083 Yang Tse (11 Jun 2009)
1084 - I adapted testcurl script to allow building test harness programs when
1085 cross-compiling for a *-*-mingw* host.
1087 Daniel Stenberg (10 Jun 2009)
1088 - Fabian Keil ran clang on the (lib)curl code, found a bunch of warnings and
1089 contributed a range of patches to fix them.
1091 Yang Tse (10 Jun 2009)
1092 - I introduced configure script option --enable-curldebug which now allows
1093 the decoupled enabling or disabling of the curl debug memory tracking
1094 feature from the --enable-debug option which no longer controls this.
1096 curl --version will list 'Debug' feature for debug enabled builds, and
1097 will list 'TrackMemory' feature for curl debug memory tracking capable
1098 builds. These features are independent and can be controlled when running
1099 the configure script. When --enable-debug is given both features will be
1100 enabled, unless some restriction prevents memory tracking from being used.
1102 Internally, definition of preprocessor symbol DEBUGBUILD restricts code
1103 which is only compiled for debug enabled builds. And symbol CURLDEBUG is
1104 used to differentiate code which is _only_ used for memory tracking.
1106 Yang Tse (9 Jun 2009)
1107 - Daniel Steinberg pointed out that Curl_FormInit() in formdata.c was not
1108 initializing the fread callback pointer and this triggered a compiler
1109 warning, also provided a friendly suggestion on how to fix it.
1111 Daniel Stenberg (8 Jun 2009)
1112 - Claes Jakobsson provided a patch for libcurl-NSS that fixed a bad refcount
1113 issue with client certs that caused issues like segfaults.
1114 http://curl.haxx.se/mail/lib-2009-05/0316.html
1116 - Triggered by bug report #2798852 and the patch in there, I fixed configure
1117 to detect gnutls build options with pkg-config only and not libgnutls-config
1118 anymore since GnuTLS has stopped distributing that tool. If an explicit path
1119 is given to configure, we will instead guess on how to link and use that
1120 lib. I did not use the patch from the bug report.
1122 Yang Tse (8 Jun 2009)
1123 - Igor Novoseltsev adjusted Makefile.vxworks to get sources and headers
1124 included from Makefile.inc, and provided docs\INSTALL VxWorks section.
1126 - I removed buildconf.bat from release and daily snapshot archives. This
1127 file is only for CVS tree checkout builds.
1129 Daniel Stenberg (8 Jun 2009)
1130 - Eric Wong fixed --no-buffer to actually switch off output buffering. Been
1133 Bill Hoffman (6 Jun 2009)
1134 - Added some cmake docs and fixed socklen_t in the build.
1136 Yang Tse (5 Jun 2009)
1137 - John E. Malmberg provided VMS specific patch: "This fixes an existing bug
1138 in urlglob.c where it was not converting the Curl Unix exit code to a VMS
1139 DCL compatible exit code. This fix required the enhancement described next.
1140 This also adds an enhancement to main.c so that when curl is run under a
1141 Unix shell like Bash on VMS, it will return the standard Unix exit codes
1142 and messages." And another patch for docs/examples.
1144 I introduced os-specific.c and os-specific.h for use in curl tool code
1145 and adjusted John E. Malmberg's patch placement to use these new files
1146 as an effort to prevent main.c from growing ad infinitum. Code already
1147 existing in main.c which is OS specific should be moved into these files.
1149 Daniel Stenberg (4 June 2009)
1150 - Setting the Content-Length: header from your app when you do a POST or PUT
1151 is almost always a VERY BAD IDEA. Yet there are still apps out there doing
1152 this, and now recently it triggered a bug/side-effect in libcurl as when
1153 libcurl sends a POST or PUT with NTLM, it sends an empty post first when it
1154 knows it will just get a 401/407 back. If the app then replaced the
1155 Content-Length header, it caused the server to wait for input that libcurl
1156 wouldn't send. Aaron Oneal reported this problem in bug report #2799008
1157 (http://curl.haxx.se/bug/view.cgi?id=2799008) and helped us verify the fix.
1159 Yang Tse (4 Jun 2009)
1160 - Igor Novoseltsev provided patches and information, that after some
1161 adjustments to better fit curl's way of doing things, have resulted
1162 in the posibility of building libcurl for VxWorks.
1164 Daniel Fandrich (2 June 2009)
1165 - Checked in a Google Android make file. To use it, you must first
1166 create a config.h file by running configure in the Android environment,
1167 which doesn't seem to be easy to do. If no easy way can be found, a
1168 static config-android.h may need to be created and checked in to the
1169 libcurl source tree.
1171 Daniel Stenberg (1 June 2009)
1172 - Claes Jakobsson fixed the configure script to better find and use NSS
1175 Yang Tse (1 Jun 2009)
1176 - John E. Malmberg provided a VMS specific clean-up for curl.h, and pointed
1177 out that the configure script was failing to detect the timeval struct on
1178 VMS when building with _XOPEN_SOURCE_EXTENDED undefined due to definition
1179 taking place in socket.h instead of time.h. I have adjusted configure
1180 script to also include this header when checking struct timeval.
1182 Daniel Stenberg (27 May 2009)
1183 - Frank McGeough provided a small OpenSSL #include fix to make libcurl compile
1184 fine with Nokia 5th edition 1.0 SDK for Symbian.
1186 - Andre Guibert de Bruet found a call to a OpenSSL function that didn't check
1187 for a failure properly.
1189 - Mike Crowe pointed out that setting CURLOPT_USERPWD to NULL used to clear
1190 the auth credentials back in 7.19.0 and earlier while now you have to set ""
1191 to get the same effect. His patch brings back the ability to use NULL.
1193 - Claes Jakobsson fixed libcurl-NSS to build fine even without the
1194 PK11_CreateGenericObject() function.
1196 Daniel Stenberg (25 May 2009)
1197 - bug report #2796358 (http://curl.haxx.se/bug/view.cgi?id=2796358) pointed
1198 out that the cookie parser would leak memory when it parses cookies that are
1199 received with domain, path etc set multiple times in the same header. While
1200 such a cookie is questionable, they occur in the wild and libcurl no longer
1201 leaks memory for them. I added such a header to test case 8.
1203 Daniel Fandrich (22 May 2009)
1204 - Removed some obsolete digest code that caused a valgrind error in test 551.
1206 Daniel Fandrich (20 May 2009)
1207 - Added "non-existing host" test keywords to make it easy to skip those
1208 tests on machines that have broken DNS configurations (such as
1209 those configured to use OpenDNS).
1211 Daniel Stenberg (19 May 2009)
1212 - Kamil Dudka brought the patch from the Redhat bug entry
1213 https://bugzilla.redhat.com/show_bug.cgi?id=427966 which was libcurl closing
1214 a bad file descriptor when closing down the FTP data connection. Caolan
1215 McNamara seems to be the original author of it.
1217 Version 7.19.5 (18 May 2009)
1219 Daniel Stenberg (17 May 2009)
1220 - James Bursa posted a patch to the mailing list that fixed a problem with
1221 no_proxy which made it not skip the proxy if the URL entered contained a
1222 user name. I added test case 1101 to verify.
1224 Daniel Stenberg (11 May 2009)
1225 - Balint Szilakszi reported a memory leak when libcurl did gzip decompression
1226 of streams that had some parts (legitimately) missing. We now provide and use
1227 a proper cleanup function for the content encoding submodule.
1228 http://curl.haxx.se/mail/lib-2009-05/0092.html
1230 - Kamil Dudka provided a fix for libcurl-NSS reported by Michael Cronenworth
1231 at https://bugzilla.redhat.com/show_bug.cgi?id=453612#c12
1233 If an incorrect password is given while loading a private key, libcurl ends
1234 up in an infinite loop consuming memory. The bug is critical.
1236 - I fixed the problem with doing NTLM, POST and then following a 302 redirect,
1237 as reported by Ebenezer Ikonne (on curl-users) and Laurent Rabret (on
1238 curl-library). The transfer was mistakenly marked to get more data to send
1239 but since it didn't actually have that, it just hung there...
1241 Daniel Stenberg (10 May 2009)
1242 - Andre Guibert de Bruet correctly pointed out an over-alloc with one wasted
1243 byte in the digest code.
1245 Yang Tse (9 May 2009)
1246 - Removed DOS and TPF package's subdirectory Makefile.am, it was only used
1247 to include some files in the distribution tarball serving no other purpose.
1248 Files from the DOS and TPF subdirectories are now included in the EXTRA_DIST
1249 of the Makefile in the parent subdirectory.
1251 Yang Tse (8 May 2009)
1252 - Changed host name literal in several tests to one under the haxx.se domain.
1254 - Renamed vc6 workspace and project files to avoid filename clash when used
1255 for conversion to later VS versions.
1257 Daniel Stenberg (8 May 2009)
1258 - Constantine Sapuntzakis fixed bug report #2784055
1259 (http://curl.haxx.se/bug/view.cgi?id=2784055) identifying a problem to
1260 connect to SOCKS proxies when using the multi interface. It turned out to
1261 almost not work at all previously. We need to wait for the TCP connect to
1262 be properly verified before doing the SOCKS magic.
1264 There's still a flaw in the FTP code for this.
1266 Daniel Stenberg (7 May 2009)
1267 - Made the SO_SNDBUF setting for the data connection socket for ftp uploads as
1268 well. See change 28 Apr 2009.
1270 Yang Tse (7 May 2009)
1271 - Fixed an issue affecting FTP transfers, introduced with the transfer.c
1272 patch committed May 4.
1274 Daniel Stenberg (7 May 2009)
1275 - Man page *roff problems fixed thanks to input from Colin Watson. Problems
1276 reported in the Debian package.
1278 - Vijay G filed bug report #2723236
1279 (http://curl.haxx.se/bug/view.cgi?id=2723236) identifying a problem with
1280 libcurl's TFTP code and its lack of dealing with the OACK packet.
1282 Yang Tse (5 May 2009)
1283 - Fixed the --ftp-port address of test #251 to the CLIENTIP address, and
1284 reverted the change affecting test suite harness committed 4 May.
1286 Daniel Stenberg (5 May 2009)
1287 - Inspired by Michael Smith's session id fix for OpenSSL, I did the
1288 corresponding fix in the GnuTLS code: make sure to store the new session id
1289 in case the previous re-used one is rejected.
1291 Daniel Stenberg (4 May 2009)
1292 - Michael Smith posted bug report #2786255
1293 (http://curl.haxx.se/bug/view.cgi?id=2786255) with a patch, identifying how
1294 libcurl did not deal with SSL session ids properly if the server rejected a
1295 re-use of one. Starting now, it will forget the rejected one and remember
1296 the new. This change was for OpenSSL only, it is likely that other SSL lib
1297 code needs similar fixes.
1299 Yang Tse (4 May 2009)
1300 - Applied David McCreedy's "transfer.c fixes for CURL_DO_LINEEND_CONV and
1301 non-ASCII platform HTTP requests" patch addressing two HTTP PUT problems:
1302 1) On non-ASCII platforms not all of the protocol portions of the PUT are
1303 being translated to ASCII. 2) On all platforms the line endings of part of
1304 the protocol portions are mangled from CRLF to CRCRLF if data->set.crlf or
1305 data->set.prefer_ascii are set (depending on CURL_DO_LINEEND_CONV).
1307 - Applied David McCreedy's patch to fix test suite harness to allow test FTP
1308 server and client on different machines, providing FTP client address when
1309 running the FTP test server.
1311 Daniel Fandrich (3 May 2009)
1312 - Added and disabled test case 563 which shows KNOWN_BUGS #59. The bug
1313 report failed to mention that a proxy must be used to reproduce it.
1315 Yang Tse (2 May 2009)
1316 - Use a build-time configured curl_socklen_t data type instead of socklen_t.
1318 Yang Tse (1 May 2009)
1319 - Applied David McCreedy's patches "TPF-platform specific changes to various
1320 files" and "http.c fix to Curl_proxyCONNECT for non-ASCII platforms", the
1321 former with minor edits.
1323 Daniel Stenberg (30 Apr 2009)
1324 - I was going to fix issue #59 in KNOWN_BUGS
1326 If the CURLOPT_PORT option is used on an FTP URL like
1327 "ftp://example.com/file;type=A" the ";type=A" is stripped off.
1329 I added test case 562 to verify, only to find out that I couldn't repeat
1330 this bug so I hereby consider it not a bug anymore!
1332 Daniel Stenberg (29 Apr 2009)
1333 - Based on bug report #2723219 (http://curl.haxx.se/bug/view.cgi?id=2723219)
1334 I've now made TFTP "connections" not being kept for re-use within libcurl.
1335 TFTP is UDP-based so the benefit was really low (if even existing) to begin
1336 with so instead of tracking down to fix this problem we instead removed the
1337 re-use. I also enabled test case 1099 that I wrote a few days ago to verify
1338 that this change fixes the reported problem.
1340 Daniel Stenberg (28 Apr 2009)
1341 - Constantine Sapuntzakis filed bug report #2783090
1342 (http://curl.haxx.se/bug/view.cgi?id=2783090) pointing out that on windows
1343 we need to grow the SO_SNDBUF buffer somewhat to get really good upload
1344 speeds. http://support.microsoft.com/kb/823764 has the details. Friends
1345 confirmed that simply adding 32 to CURL_MAX_WRITE_SIZE is enough.
1347 - Bug report #2709004 (http://curl.haxx.se/bug/view.cgi?id=2709004) by Tim
1348 Chen pointed out how curl couldn't upload with resume when reading from a
1351 This ended up with the introduction of a new return code for the
1352 CURLOPT_SEEKFUNCTION callback that basically says that the seek failed but
1353 that libcurl may try to resolve the situation anyway. In our case this means
1354 libcurl will attempt to instead read that much data from the stream instead
1355 of seeking and that way curl can now upload with resume when data is read
1358 Daniel Stenberg (26 Apr 2009)
1359 - Bug report #2779733 (http://curl.haxx.se/bug/view.cgi?id=2779733) by Sven
1360 Wegener pointed out that CURLINFO_APPCONNECT_TIME didn't work with the multi
1361 interface and provided a patch that fixed the problem!
1363 Daniel Stenberg (24 Apr 2009)
1364 - Kamil Dudka fixed another NSS-related leak when client certs were used.
1366 - Bug report #2779245 (http://curl.haxx.se/bug/view.cgi?id=2779245) by Rainer
1367 Koenig pointed out that the man page didn't tell that the *_proxy
1368 environment variables can be specified lower case or UPPER CASE and the
1369 lower case takes precedence,
1371 Daniel Fandrich (21 Apr 2009)
1372 - Added new libcurl source files to Amiga, RiscOS and VC6 build files.
1374 Yang Tse (21 Apr 2009)
1375 - Moved potential inclusion of system's malloc.h and memory.h header files to
1376 setup_once.h. Inclusion of each header file is based on the definition of
1377 NEED_MALLOC_H and NEED_MEMORY_H respectively.
1379 Renamed libcurl's memory.h to curl_memory.h
1381 Daniel Stenberg (20 Apr 2009)
1382 - Leanic Lefever reported a crash and did some detailed research on why and
1383 how it occurs (http://curl.haxx.se/mail/lib-2009-04/0289.html). The
1384 conclusion was that if an error is detected and Curl_done() is called for
1385 the connection, ftp_done() could at times return another error code that
1386 then would take precedence and that new code confused existing logic that
1387 works for the first error code (CURLE_SEND_ERROR) only.
1389 - Gisle Vanem noticed that --libtool would produce bogus strings at times for
1390 OBJECTPOINT options. Now we've introduced a new function - my_setopt_str -
1391 within the app for setting plain string options to avoid the risk of this
1394 Daniel Stenberg (17 Apr 2009)
1395 - Pramod Sharma reported and tracked down a bug when doing FTP over a HTTP
1396 proxy. libcurl would then wrongly close the connection after each
1397 request. In his case it had the weird side-effect that it killed NTLM auth
1398 for the proxy causing an inifinite loop!
1400 I added test case 1098 to verify this fix. The test case does however not
1401 properly verify that the transfers are done persistently - as I couldn't
1402 think of a clever way to achieve it right now - but you need to read the
1403 stderr output after a test run to see that it truly did the right thing.
1405 Daniel Stenberg (13 Apr 2009)
1406 - bug report #2727981 (http://curl.haxx.se/bug/view.cgi?id=2727981) by Martin
1407 Storsjö pointed out how setting CURLOPT_NOBODY to 0 could be downright
1408 confusing as it set the method to either GET or HEAD. The example he showed
1411 curl_easy_setopt(curl, CURLOPT_PUT, 1);
1412 curl_easy_setopt(curl, CURLOPT_NOBODY, 0);
1414 The new way doesn't alter the method until the request is about to start. If
1415 CURLOPT_NOBODY is then 1 the HTTP request will be HEAD. If CURLOPT_NOBODY is
1416 0 and the request happens to have been set to HEAD, it will then instead be
1417 set to GET. I believe this will be less surprising to users, and hopefully
1418 not hit any existing users badly.
1420 - Toshio Kuratomi reported a memory leak problem with libcurl+NSS that turned
1421 out to be leaking cacerts. Kamil Dudka helped me complete the fix. The issue
1422 is found in Redhat's bug tracker:
1423 https://bugzilla.redhat.com/show_bug.cgi?id=453612
1425 There are still memory leaks present, but they seem to have other reasons.
1427 Daniel Fandrich (11 Apr 2009)
1428 - Added new libcurl source files to Symbian OS build files.
1429 - Improved Symbian support for SSL.
1431 Yang Tse (10 Apr 2009)
1432 - Daniel Johnson improved the MacOSX-Framework shell script to now perform all
1433 the steps required to build a Mac OS X four way fat ppc/i386/ppc64/x86_64
1434 libcurl.framework. Four way fat framework requires OS X 10.5 SDK or later.
1436 Yang Tse (8 Apr 2009)
1437 - Removed Sun compilers preprocessor block from curlbuild.h.dist, this also
1438 removes it from the curlbuild.h file originally distributed by the cURL
1439 project as this file is intended for systems not capable of running the
1440 configure script. For those who have been building curl out of the source
1441 code curl distribution tarball provided by curl.haxx.se the change implies
1442 nothing. Previous change in this area committed 2 Apr becomes irrelevant.
1444 Daniel Stenberg (6 Apr 2009)
1445 - I clarified in the docs that CURLOPT_SEEKFUNCTION should return 0 on success
1446 and 1 on fatal errors. Previously it only mentioned non-zero on fatal
1447 errors. This is a slight change in meaning, but it follows what we've done
1448 elsewhere before and it opens up for LOTS of more useful return codes
1449 whenever we can think of them...
1451 Yang Tse (2 Apr 2009)
1452 - Fix curl_off_t definition for builds done using Sun compilers and a
1453 non-configured libcurl. In this case curl_off_t data type was gated
1454 to the off_t data type which depends on the _FILE_OFFSET_BITS. This
1455 configuration is exactly the unwanted configuration for our curl_off_t
1456 data type which must not depend on such setting. This breaks ABI for
1457 libcurl libraries built with Sun compilers which were built without
1458 having run the configure script with _FILE_OFFSET_BITS different than
1459 64 and using the ILP32 data model.
1461 Daniel Stenberg (1 Apr 2009)
1462 - Andre Guibert de Bruet fixed a NULL pointer use in an infof() call if a
1463 strdup() call failed.
1465 Daniel Fandrich (31 Mar 2009)
1466 - Properly return an error code in curl_easy_recv (reported by Jim Freeman).
1468 Daniel Stenberg (18 Mar 2009)
1469 - Kamil Dudka brought a patch that enables 6 additional crypto algorithms when
1470 NSS is used. These ciphers were added in NSS 3.4 and require to be enabled
1473 Daniel Stenberg (13 Mar 2009)
1474 - Use libssh2_version() to present the libssh2 version in case the libssh2
1475 library is found to support it.
1477 Yang Tse (12 Mar 2009)
1478 - Added missing Curl_read() return code checking in TELNET transfers.
1480 - Pierre Brico found and fixed TELNET transfers not being aborted upon
1481 a write callback failure.
1483 Daniel Stenberg (11 Mar 2009)
1484 - Kamil Dudka made the curl tool properly call curl_global_init() before any
1485 other libcurl function.
1487 Yang Tse (11 Mar 2009)
1488 - Added missing TELNET timeout support for Windows builds. This issue was
1489 reported by Pierre Brico.
1491 Daniel Stenberg (9 Mar 2009)
1492 - Frank Hempel found out a bug and provided the fix:
1494 curl_easy_duphandle did not necessarily duplicate the CURLOPT_COOKIEFILE
1495 option. It only enabled the cookie engine in the destination handle if
1496 data->cookies is not NULL (where data is the source handle). In case of a
1497 newly initialized handle which just had the cookie support enabled by a
1498 curl_easy_setopt(handle, CURL_COOKIEFILE, "")-call, handle->cookies was
1499 still NULL because the setopt-call only appends the value to
1500 data->change.cookielist, hence duplicating this handle would not have the
1501 cookie engine switched on.
1503 We also concluded that the slist-functionality would be suitable for being
1504 put in its own module rather than simply hanging out in lib/sendf.c so I
1505 created lib/slist.[ch] for them.
1507 - Andreas Farber made the 'buildconf' script check for the presence of m4
1508 scripts to make it detect a bad checkout earlier. People with older
1509 checkouts who don't do cvs update with the -d option won't get the new dirs
1510 and then will get funny outputs that can be a bit hard to understand and
1513 Daniel Stenberg (8 Mar 2009)
1514 - Andre Guibert de Bruet found and fixed a code segment in ssluse.c where the
1515 allocation of the memory BIO was not being properly checked.
1517 - Andre Guibert de Bruet fixed the gnutls-using code: There are a few places
1518 in the gnutls code where we were checking for negative values for errors,
1519 when the man pages state that GNUTLS_E_SUCCESS is returned on success and
1520 other values indicate error conditions.
1522 - Bill Egert pointed out (http://curl.haxx.se/bug/view.cgi?id=2671602) that
1523 curl didn't use sprintf() in a way that is documented to work in POSIX but
1524 since we use our own printf() code (from libcurl) that shouldn't be a
1525 problem. Nonetheless I modified the code to not rely on such particular
1526 features and to not cause further raised eyebrowse with no good reason.
1528 Daniel Fandrich (5 Mar 2009)
1529 - Expanded the security section of the libcurl-tutorial man page to cover
1530 more issues for authors to consider when writing robust libcurl-using
1533 Yang Tse (5 Mar 2009)
1534 - Fixed NTLM authentication memory leak on SSPI enabled Windows builds. This
1535 issue was noticed by Chris Deidun.
1537 Daniel Fandrich (4 Mar 2009)
1538 - Fixed a problem with m4 quoting in the OpenSSL configure check reported
1541 Daniel Stenberg (3 Mar 2009)
1542 - David James brought a patch that make libcurl close (all) dead connections
1543 whenever you attempt to open a new connection.
1545 1. After cleaning up a dead connection, "continue" instead of
1546 returning FALSE. This ensures that we clean up all dead connections,
1547 rather than just cleaning up the first dead connection.
1548 2. Move up the cleanup for dead connections so that it occurs for
1549 all connections, rather than just the connections which have the same
1550 preferences as our current new connection.
1552 Version 7.19.4 (3 March 2009)
1554 Daniel Stenberg (3 Mar 2009)
1555 - David Kierznowski notified us about a security flaw
1556 (http://curl.haxx.se/docs/adv_20090303.html also known as CVE-2009-0037) in
1557 which previous libcurl versions (by design) can be tricked to access an
1558 arbitrary local/different file instead of a remote one when
1559 CURLOPT_FOLLOWLOCATION is enabled. This flaw is now fixed in this release
1560 together this the addition of two new setopt options for controlling this
1563 o CURLOPT_REDIR_PROTOCOLS controls what protocols libcurl is allowed to
1564 follow to when CURLOPT_FOLLOWLOCATION is enabled. By default, this option
1565 excludes the FILE and SCP protocols and thus you nee to explicitly allow
1566 them in your app if you really want that behavior.
1568 o CURLOPT_PROTOCOLS controls what protocol(s) libcurl is allowed to fetch
1569 using the primary URL option. This is useful if you want to allow a user or
1570 other outsiders control what URL to pass to libcurl and yet not allow all
1571 protocols libcurl may have been built to support.
1573 Daniel Stenberg (27 Feb 2009)
1574 - Senthil Raja Velu reported a problem when CURLOPT_INTERFACE and
1575 CURLOPT_LOCALPORT were used together (the local port bind failed), and
1576 Markus Koetter provided the fix!
1578 Daniel Stenberg (25 Feb 2009)
1579 - As Daniel Fandrich figured out, we must do the GnuTLS initing in the
1580 curl_global_init() function to properly maintain the performing functions
1581 thread-safe. We've previously (28 April 2007) moved the init to a later time
1582 just to avoid it to fail very early when libgcrypt dislikes the situation,
1583 but that move was bad and the fix should rather be in libgcrypt or
1586 Daniel Stenberg (24 Feb 2009)
1587 - Brian J. Murrell found out that Negotiate proxy authentication didn't work.
1588 It happened because the code used the struct for server-based auth all the
1589 time for both proxy and server auth which of course was wrong.
1591 Daniel Stenberg (23 Feb 2009)
1592 - After a bug reported by James Cheng I've made curl_easy_getinfo() for
1593 CURLINFO_CONTENT_LENGTH_DOWNLOAD and CURLINFO_CONTENT_LENGTH_UPLOAD return
1594 -1 if the sizes aren't know. Previously these returned 0, make it impossible
1595 to detect the difference between actually zero and unknown.
1597 Yang Tse (23 Feb 2009)
1598 - Daniel Johnson provided a shell script that will perform all the steps needed
1599 to build a Mac OS X fat ppc/i386 or ppc64/x86_64 libcurl.framework
1601 Daniel Stenberg (23 Feb 2009)
1602 - I renamed everything in the windows builds files that used the name 'curllib'
1603 to the proper 'libcurl' as clearly this caused confusion.
1605 Yang Tse (20 Feb 2009)
1606 - Do not halt compilation when using VS2008 to build a Windows 2000 target.
1608 Daniel Stenberg (20 Feb 2009)
1609 - Linus Nielsen Feltzing reported and helped me repeat and fix a problem with
1610 FTP with the multi interface: when a transfer fails, like when aborted by a
1611 write callback, the control connection was wrongly closed and thus not
1614 This change is also an attempt to cleanup the code somewhat in this area, as
1615 now the FTP code attempts to keep (better) track on pending responses
1616 necessary to get read in ftp_done().
1618 Daniel Stenberg (19 Feb 2009)
1619 - Patrik Thunstrom reported a problem and helped me repeat it. It turned out
1620 libcurl did a superfluous 1000ms wait when doing SFTP downloads!
1622 We read data with libssh2 while doing the "DO" operation for SFTP and then
1623 when we were about to start getting data for the actual file part, the
1624 "TRANSFER" part, we waited for socket action (in 1000ms) before doing a
1625 libssh2-read. But in this case libssh2 had already read and buffered the
1626 data so we ended up always just waiting 1000ms before we get working on the
1629 Patrick Monnerat (18 Feb 2009)
1630 - FTP downloads (i.e.: RETR) ending with code 550 now return error
1631 CURLE_REMOTE_FILE_NOT_FOUND instead of CURLE_FTP_COULDNT_RETR_FILE.
1633 Daniel Stenberg (17 Feb 2009)
1634 - Kamil Dudka made NSS-powered builds compile and run again!
1636 - A second follow-up change by Andre Guibert de Bruet to fix a related memory
1637 leak like that fixed on the 14th. When zlib returns failure, we need to
1638 cleanup properly before returning error.
1640 - CURLOPT_FTP_CREATE_MISSING_DIRS can now be set to 2 in addition to 1 for
1641 plain FTP connections, and it will then allow MKD to fail once and retry the
1642 CWD afterwards. This is especially useful if you're doing many simultanoes
1643 connections against the same server and they all have this option enabled,
1644 as then CWD may first fail but then another connection does MKD before this
1645 connection and thus MKD fails but trying CWD works! The numbers can
1646 (should?) now be set with the convenience enums now called
1647 CURLFTP_CREATE_DIR and CURLFTP_CREATE_DIR_RETRY.
1649 Tests has proven that if you're making an application that uploads a set of
1650 files to an ftp server, you will get a noticable gain in speed if you're
1651 using multiple connections and this option will be then be very useful.
1653 Daniel Stenberg (14 Feb 2009)
1654 - Andre Guibert de Bruet found and fixed a memory leak in the content encoding
1655 code, which could happen on libz errors.
1657 Daniel Fandrich (12 Feb 2009)
1658 - Added support for Digest and NTLM authentication using GnuTLS.
1660 Daniel Stenberg (11 Feb 2009)
1661 - CURLINFO_CONDITION_UNMET was added to allow an application to get to know if
1662 the condition in the previous request was unmet. This is typically a time
1663 condition set with CURLOPT_TIMECONDITION and was previously not possible to
1664 reliably figure out. From bug report #2565128
1665 (http://curl.haxx.se/bug/view.cgi?id=2565128) filed by Jocelyn Jaubert.
1667 Daniel Fandrich (4 Feb 2009)
1668 - Don't add the standard /usr/lib or /usr/include paths to LDFLAGS and CPPFLAGS
1669 (respectively) when --with-ssl=/usr is used (patch based on FreeBSD).
1671 - Added an explicit buffer limit check in msdosify() (patch based on FreeBSD).
1672 This couldn't ever overflow in curl, but might if the code were used
1673 elsewhere or under different conditions.
1675 Daniel Stenberg (3 Feb 2009)
1676 - Hidemoto Nakada provided a small fix that makes it possible to get the
1677 CURLINFO_CONTENT_LENGTH_DOWNLOAD size from file:// "transfers" with
1678 CURLOPT_NOBODY set true.
1680 Daniel Stenberg (2 Feb 2009)
1681 - Patrick Scott found a rather large memory leak when using the multi
1682 interface and setting CURLMOPT_MAXCONNECTS to something less than the number
1683 of handles you add to the multi handle. All the connections that didn't fit
1684 in the cache would not be properly disconnected nor freed!
1686 - Craig A West brought us: libcurl now defaults to do CONNECT with HTTP
1687 version 1.1 instead of 1.0 like before. This change also introduces the new
1688 proxy type for libcurl called 'CURLPROXY_HTTP_1_0' that then allows apps to
1689 switch (back) to CONNECT 1.0 requests. The curl tool also got a --proxy1.0
1690 option that works exactly like --proxy but sets CURLPROXY_HTTP_1_0.
1692 I updated all test cases cases that use CONNECT and I tried to do some using
1693 --proxy1.0 and some updated to do CONNECT 1.1 to get both versions run.
1695 Daniel Stenberg (31 Jan 2009)
1696 - When building with c-ares 1.6.1 (not yet released) or later and IPv6 support
1697 enabled, we can now take advantage of its brand new AF_UNSPEC support in
1698 ares_gethostbyname(). This makes test case 241 finally run fine for me with
1699 this setup since it now parses the "::1 ip6-localhost" line fine in my
1702 Daniel Stenberg (30 Jan 2009)
1703 - Scott Cantor filed bug report #2550061
1704 (http://curl.haxx.se/bug/view.cgi?id=2550061) mentioning that I failed to
1705 properly make sure that the VC9 makefiles got included in the latest
1706 release. I've now fixed the release script and verified it so next release
1707 will hopefully include them properly!
1709 Daniel Fandrich (30 Jan 2009)
1710 - Fixed --disable-proxy for FTP and SOCKS. Thanks to Daniel Egger for
1713 Yang Tse (29 Jan 2009)
1714 - Introduced curl_sspi.c and curl_sspi.h for the implementation of functions
1715 Curl_sspi_global_init() and Curl_sspi_global_cleanup() which previously were
1716 named Curl_ntlm_global_init() and Curl_ntlm_global_cleanup() in http_ntlm.c
1717 Also adjusted socks_sspi.c to remove the link-time dependency on the Windows
1718 SSPI library using it now in the same way as it was done in http_ntlm.c.
1720 Daniel Stenberg (28 Jan 2009)
1721 - Markus Moeller introduced two new options to libcurl:
1722 CURLOPT_SOCKS5_GSSAPI_SERVICE and CURLOPT_SOCKS5_GSSAPI_NEC to allow libcurl
1723 to do GSS-style authentication with SOCKS5 proxies. The curl tool got the
1724 options called --socks5-gssapi-service and --socks5-gssapi-nec to enable
1727 Daniel Stenberg (26 Jan 2009)
1728 - Chad Monroe provided the new CURLOPT_TFTP_BLKSIZE option that allows an app
1729 to set desired block size to use for TFTP transfers instead of the default
1732 - The "-no_ticket" option was introduced in Openssl0.9.8j. It's a flag to
1733 disable "rfc4507bis session ticket support". rfc4507bis was later turned
1734 into the proper RFC5077 it seems: http://tools.ietf.org/html/rfc5077
1736 The enabled extension concerns the session management. I wonder how often
1737 libcurl stops a connection and then resumes a TLS session. also, sending the
1738 session data is some overhead. .I suggest that you just use your proposed
1739 patch (which explicitly disables TICKET).
1741 If someone writes an application with libcurl and openssl who wants to
1742 enable the feature, one can do this in the SSL callback.
1744 Sharad Gupta brought this to my attention. Peter Sylvester helped me decide
1745 on the proper action.
1747 - Alexey Borzov filed bug report #2535504
1748 (http://curl.haxx.se/bug/view.cgi?id=2535504) pointing out that realms with
1749 quoted quotation marks in HTTP Digest headers didn't work. I've now added
1750 test case 1095 that verifies my fix.
1752 - Craig A West brought CURLOPT_NOPROXY and the corresponding --noproxy option.
1753 They basically offer the same thing the NO_PROXY environment variable only
1754 offered previously: list a set of host names that shall not use the proxy
1755 even if one is specified.
1757 Daniel Fandrich (20 Jan 2009)
1758 - Call setlocale() for libtest tests to test the effects of locale-induced
1759 libc changes on libcurl.
1761 - Fixed a couple more locale-dependent toupper conversions, mainly for
1762 clarity. This does fix one problem that causes ;type=i FTP URLs
1763 to fail in the Turkish locale when CURLOPT_PROXY_TRANSFER_MODE is
1764 used (test case 561)
1766 - Added tests 561 and 1091 through 1094 to test various combinations
1767 of ;type= and ;mode= URLs that could potentially fail in the Turkish
1770 Daniel Stenberg (20 Jan 2009)
1771 - Lisa Xu pointed out that the ssh.obj file was missing from the
1772 lib/Makefile.vc6 file (and thus from the vc8 and vc9 ones too).
1774 Version 7.19.3 (19 January 2009)
1776 Daniel Stenberg (16 Jan 2009)
1777 - Andrew de los Reyes fixed curlbuild.h for "generic" gcc builds on PPC, both
1780 Daniel Stenberg (15 Jan 2009)
1781 - Tim Ansell fixed a compiler warning in lib/cookie.c
1783 Daniel Stenberg (14 Jan 2009)
1784 - Grant Erickson fixed timeouts for TFTP such that specifying a
1785 connect-timeout, a max-time or both options work correctly and as expected
1786 by passing the correct boolean value to Curl_timeleft via the
1787 'duringconnect' parameter.
1789 With this small change, curl TFTP now behaves as expected (and likely as
1790 originally-designed):
1792 1) For non-existent or unreachable dotted IP addresses:
1794 a) With no options, follows the default curl 300s timeout...
1795 b) With --connect-timeout only, follows that value...
1796 c) With --max-time only, follows that value...
1797 d) With both --connect-timeout and --max-time, follows the smaller value...
1799 and times out with a "curl: (7) Couldn't connect to server" error.
1801 2) For transfers to/from a valid host:
1803 a) With no options, follows default curl 300s timeout for the
1804 first XRQ/DATA/ACK transaction and the default TFTP 3600s
1805 timeout for the remainder of the transfer...
1807 b) With --connect-time only, follows that value for the
1808 first XRQ/DATA/ACK transaction and the default TFTP 3600s
1809 timeout for the remainder of the transfer...
1811 c) With --max-time only, follows that value for the first
1812 XRQ/DATA/ACK transaction and for the remainder of the
1815 d) With both --connect-timeout and --max-time, follows the former
1816 for the first XRQ/DATA/ACK transaction and the latter for the
1817 remainder of the transfer...
1819 and times out with a "curl: (28) Timeout was reached" error as
1822 Daniel Stenberg (13 Jan 2009)
1823 - Michael Wallner fixed a NULL pointer deref when calling
1824 curl_easy_setup(curl, CURLOPT_COOKIELIST, "SESS") on a CURL handle with no
1827 - Stefan Teleman brought a patch to fix the default curlbuild.h file for the
1830 Daniel Stenberg (12 Jan 2009)
1831 - Based on bug report #2498665 (http://curl.haxx.se/bug/view.cgi?id=2498665)
1832 by Daniel Black, I've now added magic to the configure script that makes it
1833 use pkg-config to detect gnutls details as well if the existing method
1834 (using libgnutls-config) fails. While doing this, I cleaned up and unified
1835 the pkg-config usage when detecting openssl and nss as well.
1837 Daniel Stenberg (11 Jan 2009)
1838 - Karl Moerder brought the patch that creates vc9 Makefiles, and I made
1839 'maketgz' now use the actual makefile targets to do the VC8 and VC9
1842 Daniel Stenberg (10 Jan 2009)
1843 - Emil Romanus fixed:
1845 When using the multi interface over HTTP and the server returns a Location
1846 header, the running easy handle will get stuck in the CURLM_STATE_PERFORM
1847 state, leaving the external event loop stuck waiting for data from the
1848 ingoing socket (when using the curl_multi_socket_action stuff). While this
1849 bug was pretty hard to find, it seems to require only a one-line fix. The
1850 break statement on line 1374 in multi.c caused the function to skip the call
1853 How to reproduce this bug? Well, that's another question. evhiperfifo.c in
1854 the examples directory chokes on this bug only _sometimes_, probably
1855 depending on how fast the URLs are added. One way of testing the bug out is
1856 writing to hiper.fifo from more than one source at the same time.
1858 Daniel Fandrich (7 Jan 2009)
1859 - Unified much of the SessionHandle initialization done in Curl_open() and
1860 curl_easy_reset() by creating Curl_init_userdefined(). This had the side
1861 effect of fixing curl_easy_reset() so it now also resets
1862 CURLOPT_FTP_FILEMETHOD and CURLOPT_SSL_SESSIONID_CACHE
1864 Daniel Stenberg (7 Jan 2009)
1865 - Rob Crittenden did once again provide an NSS update:
1867 I have to jump through a few hoops now with the NSS library initialization
1868 since another part of an application may have already initialized NSS by the
1869 time Curl gets invoked. This patch is more careful to only shutdown the NSS
1870 library if Curl did the initialization.
1872 It also adds in a bit of code to set the default ciphers if the app that
1873 call NSS_Init* did not call NSS_SetDomesticPolicy() or set specific
1874 ciphers. One might argue that this lets other application developers get
1875 lazy and/or they aren't using the NSS API correctly, and you'd be right.
1876 But still, this will avoid terribly difficult-to-trace crashes and is
1879 Daniel Stenberg (1 Jan 2009)
1880 - 'reconf' is removed since we rather have users use 'buildconf'
1882 Daniel Stenberg (31 Dec 2008)
1883 - Bas Mevissen reported http://curl.haxx.se/bug/view.cgi?id=2479030 pointing
1884 out that 'reconf' didn't properly point out the m4 subdirectory when running
1887 Daniel Stenberg (29 Dec 2008)
1888 - Phil Lisiecki filed bug report #2413067
1889 (http://curl.haxx.se/bug/view.cgi?id=2413067) that identified a problem that
1890 would cause libcurl to mark a DNS cache entry "in use" eternally if the
1891 subsequence TCP connect failed. It would thus never get pruned and refreshed
1892 as it should've been.
1894 Phil provided his own patch to this problem that while it seemed to work
1895 wasn't complete and thus I wrote my own fix to the problem.
1897 Daniel Stenberg (28 Dec 2008)
1898 - Peter Korsgaard fixed building libcurl with "configure --with-ssl
1901 - Anthony Bryan fixed more language and spelling flaws in man pages.
1903 Daniel Stenberg (22 Dec 2008)
1904 - Given a recent enough libssh2, libcurl can now seek/resume with SFTP even
1905 on file indexes beyond 2 or 4GB.
1907 - Anthony Bryan provided a set of patches that cleaned up manual language,
1908 corrected spellings and more.
1910 Daniel Stenberg (20 Dec 2008)
1911 - Igor Novoseltsev fixed a bad situation for the multi_socket() API when doing
1912 pipelining, as libcurl could then easily get confused and A) work on the
1913 handle that was not "first in queue" on a pipeline, or even B) tell the app
1914 to REMOVE a socket while it was in use by a second handle in a pipeline. Both
1915 errors caused hanging or stalling applications.
1917 Daniel Stenberg (19 Dec 2008)
1918 - curl_multi_timeout() could return a timeout value of 0 even though nothing
1919 was actually ready to get done, as the internal time resolution is higher
1920 than the returned millisecond timer. Therefore it could cause applications
1921 running on fast processors to do short bursts of busy-loops.
1922 curl_multi_timeout() will now only return 0 if the timeout is actually
1925 - Using the libssh2 0.19 function libssh2_session_block_directions(), libcurl
1926 now has an improved ability to do right when the multi interface (both
1927 "regular" and multi_socket) is used for SCP and SFTP transfers. This should
1928 result in (much) less busy-loop situations and thus less CPU usage with no
1931 Daniel Stenberg (17 Dec 2008)
1932 - SCP and SFTP with the multi interface had the same flaw: the 'DONE'
1933 operation didn't complete properly if the EAGAIN equivalent was returned but
1934 libcurl would simply continue with a half-completed close operation
1935 performed. This ruined persistent connection re-use and cause some
1936 SSH-protocol errors in general. The correction is unfortunately adding a
1937 blocking function - doing it entirely non-blocking should be considered for
1940 Gisle Vanem (16 Dec 2008)
1941 - Added the possibility to use the Watt-32 tcp/ip stack under Windows.
1942 The change simply involved adding a USE_WATT32 section in the
1943 config-win32.h files (under ./lib and ./src). This section disables
1944 the use of any Winsock headers.
1946 Daniel Stenberg (16 Dec 2008)
1947 - libssh2_sftp_last_error() was wrongly used at some places in libcurl which
1948 made libcurl sometimes not properly abort problematic SFTP transfers.
1950 Daniel Stenberg (12 Dec 2008)
1951 - More work with Igor Novoseltsev to first fix the remaining stuff for
1952 removing easy handles from multi handles when the easy handle is/was within
1953 a HTTP pipeline. His bug report #2351653
1954 (http://curl.haxx.se/bug/view.cgi?id=2351653) was also related and was
1955 eventually fixed by a patch by Igor himself.
1957 Yang Tse (12 Dec 2008)
1958 - Patrick Monnerat fixed a build regression, introduced in 7.19.2, affecting
1959 OS/400 compilations with IPv6 enabled.
1961 Daniel Stenberg (12 Dec 2008)
1962 - Mark Karpeles filed bug report #2416182 titled "crash in ConnectionExists
1963 when using duphandle+curl_mutli"
1964 (http://curl.haxx.se/bug/view.cgi?id=2416182) which showed that
1965 curl_easy_duphandle() wrongly also copied the pointer to the connection
1966 cache, which was plain wrong and caused a segfault if the handle would be
1967 used in a different multi handle than the handle it was duplicated from.
1969 Daniel Stenberg (11 Dec 2008)
1970 - Keshav Krity found out that libcurl failed to deal with dotted IPv6
1971 addresses if they were very long (>39 letters) due to a too strict address
1972 validity parser. It now accepts addresses up to 45 bytes long.
1974 Daniel Stenberg (11 Dec 2008)
1975 - Internet Explorer had a broken HTTP digest authentication before v7 and
1976 there are servers "out there" that relies on the client doing this broken
1977 Digest authentication. Apache even comes with an option to work with such
1980 The difference is only for URLs that contain a query-part (a '?'-letter and
1981 text to the right of it).
1983 libcurl now supports this quirk, and you enable it by setting the
1984 CURLAUTH_DIGEST_IE bit in the bitmask you pass to the CURLOPT_HTTPAUTH or
1985 CURLOPT_PROXYAUTH options. They are thus individually controlled to server
1988 (note that there's no way to activate this with the curl tool yet)
1990 Daniel Fandrich (9 Dec 2008)
1991 - Added test cases 1089 and 1090 to test --write-out after a redirect to
1992 test a report that the size didn't work, but these test cases pass.
1994 - Documented CURLOPT_CONNECT_ONLY as being useful only on HTTP URLs.
1996 Daniel Stenberg (9 Dec 2008)
1997 - Ken Hirsch simplified how libcurl does FTPS: now it doesn't assume any
1998 particular state for the control connection like it did before for implicit
1999 FTPS (libcurl assumed such control connections to be encrypted while some
2000 FTPS servers such as FileZilla assumes such connections to be clear
2001 mode). Use the CURLOPT_USE_SSL option to set your desired level.
2003 Daniel Stenberg (8 Dec 2008)
2004 - Fred Machado posted about a weird FTP problem on the curl-users list and when
2005 researching it, it turned out he got a 550 response back from a SIZE command
2006 and then I fell over the text in RFC3659 that says:
2008 The presence of the 550 error response to a SIZE command MUST NOT be taken
2009 by the client as an indication that the file cannot be transferred in the
2010 current MODE and TYPE.
2012 In other words: the change I did on September 30th 2008 and that has been
2013 included in the last two releases were a regression and a bad idea. We MUST
2014 NOT take a 550 response from SIZE as a hint that the file doesn't exist.
2016 - Christian Krause filed bug #2221237
2017 (http://curl.haxx.se/bug/view.cgi?id=2221237) that identified an infinite
2018 loop during GSS authentication given some specific conditions. With his
2019 patience and great feedback I managed to narrow down the problem and
2020 eventually fix it although I can't test any of this myself!
2022 Daniel Fandrich (3 Dec 2008)
2023 - Fixed the getifaddrs version of Curl_if2ip to work on systems without IPv6
2024 support (e.g. Minix)
2026 Daniel Stenberg (3 Dec 2008)
2027 - Igor Novoseltsev filed bug #2351645
2028 (http://curl.haxx.se/bug/view.cgi?id=2351645) that identified a problem with
2029 the multi interface that occured if you removed an easy handle while in
2030 progress and the handle was used in a HTTP pipeline.
2032 - Pawel Kierski pointed out a mistake in the cookie code that could lead to a
2033 bad fclose() after a fatal error had occured.
2034 (http://curl.haxx.se/bug/view.cgi?id=2382219)
2036 Daniel Fandrich (25 Nov 2008)
2037 - If a HTTP request is Basic and num is already >=1000, the HTTP test
2038 server adds 1 to num to get the data section to return. This allows
2039 testing authentication negotiations using the Basic authentication
2042 - Added tests 1087 and 1088 to test Basic authentication on a redirect
2043 with and without --location-trusted
2045 Daniel Stenberg (24 Nov 2008)
2046 - Based on a patch by Vlad Grachov, libcurl now uses a new libssh2 0.19
2047 function when built to support SCP and SFTP that helps the library to know
2048 in which direction a particular libssh2 operation would return EAGAIN so
2049 that libcurl knows what socket conditions to wait for before trying the
2050 function call again. Previously (and still when using libssh2 0.18 or
2051 earlier), libcurl will busy-loop in this situation when the easy interface
2054 Daniel Fandrich (20 Nov 2008)
2055 - Automatically detect OpenBSD's CA cert bundle.
2057 Daniel Stenberg (19 Nov 2008)
2058 - I removed the default use of "Pragma: no-cache" from libcurl when a proxy is
2059 used. It has been used since forever but it was never a good idea to use
2060 unless explicitly asked for.
2062 - Josef Wolf's extension that allows a $TESTDIR/gdbinit$testnum file that when
2063 you use runtests.pl -g, will be sourced by gdb to allow additional fancy or
2064 whatever you see fit
2066 - Christian Krause reported and fixed a memory leak that would occur with HTTP
2067 GSS/kerberos authentication (http://curl.haxx.se/bug/view.cgi?id=2284386)
2069 - Andreas Wurf and Markus Koetter helped me analyze a problem that Andreas got
2070 when uploading files to a single FTP server using multiple easy handle
2071 handles with the multi interface. Occasionally a handle would stall in
2074 The problem turned out to be a side-effect of the ConnectionExists()
2075 function's eagerness to re-use a handle for HTTP pipelining so it would
2076 select it even if already being in use, due to an inadequate check for its
2077 chances of being used for pipelnining.
2079 Daniel Fandrich (17 Nov 2008)
2080 - Added more compiler warning options for gcc 4.3
2082 Yang Tse (17 Nov 2008)
2083 - Fix a remaining problem in the inet_pton() runtime configure check. And
2084 fix internal Curl_inet_pton() failures to reject certain malformed literals.
2086 - Make configure script check if ioctl with the SIOCGIFADDR command can be
2087 used, and define HAVE_IOCTL_SIOCGIFADDR if appropriate.
2089 Daniel Stenberg (16 Nov 2008)
2090 - Christian Krause fixed a build failure when building with gss support
2091 enabled and FTP disabled.
2093 - Added check for NULL returns from strdup() in src/main.c and lib/formdata.c
2094 - reported by Jim Meyering also prevent buffer overflow on MSDOS when you do
2095 for example -O on a url with a file name part longer than PATH_MAX letters
2097 - lib/nss.c fixes based on the report by Jim Meyering: I went over and added
2098 checks for return codes for all calls to malloc and strdup that were
2099 missing. I also changed a few malloc(13) to use arrays on the stack and a
2100 few malloc(PATH_MAX) to instead use aprintf() to lower memory use.
2102 - I fixed a memory leak in Curl_nss_connect() when CURLOPT_ISSUERCERT is
2105 Daniel Fandrich (14 Nov 2008)
2106 - Added .xml as one of the few common file extensions known by the multipart
2109 - Added some #ifdefs around header files and change the EAGAIN test to
2110 fix compilation on Cell (reported by Jeff Curley).
2112 Yang Tse (14 Nov 2008)
2113 - Fixed several configure script issues affecting checks for inet_ntoa_r(),
2114 inet_ntop(), inet_pton(), getifaddrs(), fcntl() and getaddrinfo().
2116 Yang Tse (13 Nov 2008)
2117 - Refactored configure script detection of functions used to set sockets into
2118 non-blocking mode, and decouple function detection from function capability.