1 _ _ ____ _ 2 ___| | | | _ \| | 3 / __| | | | |_) | | 4 | (__| |_| | _ <| |___ 5 \___|\___/|_| \_\_____| 6 7 Changelog 8 9Version 7.21.0 (16 June 2010) 10 11Daniel Stenberg (5 June 2010) 12- Constantine Sapuntzakis fixed a case of spurious SSL connection aborts using 13 libcurl and OpenSSL. "I tracked it down to uncleared error state on the 14 OpenSSL error stack - patch attached deals with that." 15 16Daniel Stenberg (5 June 2010) 17- Frank Meier added CURLINFO_PRIMARY_PORT, CURLINFO_LOCAL_IP and 18 CURLINFO_LOCAL_PORT to curl_easy_getinfo(). 19 20Yang Tse (4 June 2010) 21- Enabled OpenLDAP support for cygwin builds. This support was disabled back 22 in 2008 due to incompatibilities between OpenSSL and OpenLDAP headers. 23 cygwin's OpenSSL 0.9.8l and OpenLDAP 2.3.43 versions on cygwin 1.5.25 24 allow building an OpenLDAP enabled libcurl supporting back to Windows 95. 25 26 Removed the non-functional CURL_LDAP_HYBRID code and references. 27 28Daniel Stenberg (2 June 2010) 29- Jason McDonald posted bug report #3006786 when he found that the SFTP code 30 didn't timeout properly in several places in the code even if a timeout was 31 set properly. 32 33 Based on his suggested patch, I wrote a different implementation that I 34 think addressed the issue better and also uses the connect timeout for the 35 initial part of the SSH/SFTP done during the "protocol connect" phase. 36 37 (http://curl.haxx.se/bug/view.cgi?id=3006786) 38 39Yang Tse (2 June 2010) 40- Added missing new libcurl files to non-configure targets. Adjusted 41 libcurl standard internal header inclusions in new files. Fixed an 42 SPNEGO related memory leak. Fixed several LDAP related compilation 43 issues, and fixed some compiler warnings. 44 45Daniel Stenberg (1 June 2010) 46- Igor Novoseltsev reported a problem with the multi socket API and using 47 timeouts and timers. It boiled down to a problem with libcurl's use of 48 GetTickCount() interally to figure out the current time, while Igor's own 49 application code used another function call. 50 51 It made his app call the socket API timeout function a bit _before_ libcurl 52 would consider the timeout to trigger, and that could easily lead to 53 timeouts or stalls in the app. It seems GetTickCount() in general often has 54 no better resolution than 16ms and switching to the alternative function 55 QueryPerformanceCounter has its share of problems: 56 http://www.virtualdub.org/blog/pivot/entry.php?id=106 57 58 We address this problem by simply having libcurl treat timers that already 59 has occured or will occur within 40ms subject for treatment. I'm confident 60 that there are other implementations and operating systems with similarly in 61 accurate timer functions so it makes sense to have applied generically and I 62 don't believe we sacrifice much by adding a 40ms inaccuracy on these 63 timeouts. 64 65Kamil Dudka (27 May 2010) 66- added a new test for CRL support (test313) 67 68- Tor Arntsen changed the alternative definition of bool to use enum instead 69 of unsigned char. 70 71Daniel Stenberg (25 May 2010) 72- Julien Chaffraix fixed the warning seen when compiling lib/rtmp.c: one 73 unused variables, several unused arguments and some missing #include. 74 75- Julien Chaffraix fixed 2 OOM errors: a missing NULL-check in 76 lib/http_negociate.c and a potential NULL dereferencing in lib/splay.c 77 78- Howard Chu brought a patch that makes the LDAP code much cleaner, nicer and 79 in general being a better libcurl citizen. If a new enough OpenLDAP version 80 is detect, the new and shiny lib/openldap.c code is then used instead of the 81 old cruft. 82 83Daniel Stenberg (21 May 2010) 84- Eric Mertens posted bug #3003705: when we made TFTP use the correct timeout 85 option when sent to the server (fixed May 18th 2010) it became obvious that 86 libcurl used invalid timeout values (300 by default while the RFC allows 87 nothing above 255). While of course it is obvious that as TFTP has worked 88 thus far without being able to set timeout at all, just removing the setting 89 wouldn't make any difference in behavior. I decided to still keep it (but 90 fix the problem) as it now actually allows for easier (future) customization 91 of the timeout. 92 93 (http://curl.haxx.se/bug/view.cgi?id=3003705) 94 95- Douglas Kilpatrick filed bug report #3004787 and pointed out that the TFTP 96 code didn't handle block id wraps correctly. His suggested fix inspired the 97 fix I committed. 98 99 (http://curl.haxx.se/bug/view.cgi?id=3004787) 100 101Daniel Stenberg (20 May 2010) 102- Tanguy Fautre brought a fix to allow curl to build with Microsoft VC10. 103 104Daniel Stenberg (18 May 2010) 105- Eric Mertens posted bug report #3003005 pointing out that the libcurl TFTP 106 code was not sending the timeout option properly to the server, and 107 suggested a fix. 108 109 (http://curl.haxx.se/bug/view.cgi?id=3003005) 110 111Kamil Dudka (16 May 2010) 112- Pavel Raiskup introduced a new option CURLOPT_FNMATCH_DATA in order to pass 113 a custom data pointer to the callback specified by CURLOPT_FNMATCH_FUNCTION. 114 115Daniel Stenberg (14 May 2010) 116- John-Mark Bell filed bug #3000052 that identified a problem (with an 117 associated patch) with the OpenSSL handshake state machine when the multi 118 interface is used: 119 120 Performing an https request using a curl multi handle and using select or 121 epoll to wait for events results in a hang. It appears that the cause is the 122 fix for bug #2958179, which makes ossl_connect_common unconditionally return 123 from the step 2 loop when fetching from a multi handle. 124 125 When ossl_connect_step2 has completed, it updates connssl->connecting_state 126 to ssl_connect_3. ossl_connect_common will then return to the caller, as a 127 multi handle is in use. Eventually, the client code will call 128 curl_multi_fdset to obtain an updated fdset to select or epoll on. For https 129 requests, curl_multi_fdset will cause https_getsock to be called. 130 https_getsock will only return a socket handle if the connecting_state is 131 ssl_connect_2_reading or ssl_connect_2_writing. Therefore, the client will 132 never obtain a valid fdset, and thus not drive the multi handle, resulting 133 in a hang. 134 135 (http://curl.haxx.se/bug/view.cgi?id=3000052) 136 137- Sebastian V reported bug #3000056 identifying a problem with redirect 138 following. It showed that when curl followed redirects it didn't properly 139 ignore the response body of the 30X response if that response was using 140 compressed Content-Encoding! 141 142 (http://curl.haxx.se/bug/view.cgi?id=3000056) 143 144Daniel Stenberg (12 May 2010) 145- Howard Chu brought support for RTMP. This is powered by the underlying 146 librtmp library. It supports a range of variations and "sub-protocols" 147 within the RTMP family. 148 149- Pavel Raiskup brought support for FTP directory wildcard matching to allow 150 selective downloading. To provide that, a set of new options were added: 151 152 CURLOPT_WILDCARDMATCH 153 CURLOPT_CHUNK_BGN_FUNCTION 154 CURLOPT_CHUNK_END_FUNCTION 155 CURLOPT_CHUNK_DATA 156 CURLOPT_FNMATCH_FUNCTION 157 158 There were also a set of new tests added (574 - 577) to verify this. 159 160Kamil Dudka (11 May 2010) 161- CRL support in libcurl-NSS has been completely broken. Now it works. Original 162 bug report: https://bugzilla.redhat.com/581926 163 164Daniel Stenberg (7 May 2010) 165- Dirk Manske reported a regression. When connecting with the multi interface, 166 there were situations where libcurl wouldn't store connect time correctly as 167 it used to (and is documented to) do. 168 169 Using his fine sample program we could repeat it, and I wrote up test case 170 573 using that code. The problem does not easily show itself using the local 171 test suite though. 172 173 The fix, also as suggested by Dirk, is a bit on the ugly side as it adds yet 174 another call to Curl_verboseconnect() and setting the TIMER_CONNECT time. 175 That situation is subject for some closer inspection in the future. 176 177- Howard Chu split the I/O handling functions into private handlers. 178 179 Howard Chu brought the bulk work of this patch that properly moves out the 180 sending and recving of data to the parts of the code that are properly 181 responsible for the various ways of doing so. 182 183 Daniel Stenberg assisted with polishing a few bits and fixed some minor 184 flaws in the original patch. 185 186 Another upside of this patch is that we now abuse CURLcodes less with the 187 "magic" -1 return codes and instead use CURLE_AGAIN more consistently. 188 189Daniel Stenberg (5 May 2010) 190- Hoi-Ho Chan introduced support for using the PolarSSL library. You control 191 this with the new configure option --with-polarssl. 192 193Daniel Stenberg (29 Apr 2010) 194- Ben Greear made telnet a lot better/easier to use by an application: 195 196 The main change is to allow input from user-specified methods, when they are 197 specified with CURLOPT_READFUNCTION. All calls to fflush(stdout) in 198 telnet.c were removed, which makes using 'curl telnet://foo.com' painful 199 since prompts and other data are not always returned to the user promptly. 200 Use 'curl --no-buffer telnet://foo.com' instead. In general, the user 201 should have their CURLOPT_WRITEFUNCTION do a fflush for interactive use. 202 203 Also fix assumption that reading from stdin never returns < 0. 204 Old code could crash in that case. 205 206 Call progress functions in telnet main loop. 207 208Daniel Stenberg (26 Apr 2010) 209- Make use of the libssh2_init/exit functions that libssh2 added in version 210 1.2.5. Using them will improve how libcurl works in threaded situations when 211 SCP and SFTP are transfered. 212 213Daniel Stenberg (25 Apr 2010) 214- Based on work by Kamil Dudka, I've introduced the new configure option 215 --enable-threaded-resolver. When used, the configure script will check for 216 pthreads and if around, it will build libcurl to use pthreads to do name 217 resolving in a threaded manner. Note that this is just a fix to offer an 218 option that can enable the code that already included. The threader resolver 219 code was mostly added on Jan 26 2010. 220 221Daniel Stenberg (24 Apr 2010) 222- Alex Bligh introduced the --proto and -proto-redir options that limit what 223 protocols curl accepts for the requests and when following redirects. 224 225Kamil Dudka (24 Apr 2010) 226- Fixed test536 in order to not fail with threaded DNS resolver and tweaked 227 comments in certain examples using curl_multi_fdset(). 228 229- Fixed SSL handshake timeout underflow in libcurl-NSS, which caused test405 230 to hang on a slow machine. 231 232Daniel Stenberg (21 Apr 2010) 233- The -O option caused curl to crash on windows and DOS due to the tool 234 writing out of boundary memory. 235 236Yang Tse (20 Apr 2010) 237- Ruslan Gazizov detected that MSVC makefiles were using wsock32.lib instead 238 of ws2_32.lib, this generated linking issues on MSVC IPv6 enabled builds 239 that were done using those makefiles. 240 241Daniel Stenberg (19 Apr 2010) 242- -J/--remote-header-name didn't strip trailing carriage returns or linefeeds 243 properly, so they could be used in the file name. 244 245Daniel Stenberg (16 Apr 2010) 246- Jerome Vouillon made the GnuTLS SSL handshake phase non-blocking. 247 248- The recent overhaul of the SSL recv function made the GnuTLS specific code 249 treat a zero returned from gnutls_record_recv() as an error, and this caused 250 our HTTPS test cases to fail. We leave it to upper layer code to detect if 251 an EOF is a problem or not. 252 253- I reverted the resolver fix from yesterday and instead removed all uses of 254 AI_CANONNAME all over libcurl and made the only user of that info (krb5.c) 255 use the host name from the URL instead. No reverse resolving is a good 256 thing. 257 258- Paul Howarth made configure properly detect GSS "on ancient Linux distros" 259 by editing in which order we use headers to detect GSS. 260 261Daniel Stenberg (15 Apr 2010) 262- Rainer Canavan filed bug report #2987196 that identified libcurl doing 263 unnecesary reverse name lookups in many cases when built to use IPv4 and 264 getaddrinfo(). The logic for ipv6 is now used for ipv4 too. 265 266 (http://curl.haxx.se/bug/view.cgi?id=2963679) 267 268Version 7.20.1 (14 April 2010) 269 270Daniel Stenberg (9 Apr 2010) 271- Prefixing the FTP quote commands with an asterisk really only worked for the 272 postquote actions. This is now fixed and test case 227 has been extended to 273 verify. 274 275Kamil Dudka (4 Apr 2010) 276- Eliminated a race condition in Curl_resolv_timeout(). 277 278- Refactorized interface of Curl_ssl_recv()/Curl_ssl_send(). 279 280- libcurl-NSS now provides more accurate messages and error codes in case of 281 client certificate problem. Either during connection, or transfer phase. 282 283Daniel Stenberg (1 Apr 2010) 284- Matt Wixson found and fixed a bug in the SCP/SFTP area where the code 285 treated a 0 return code from libssh2 to be the same as EAGAIN while in 286 reality it isn't. The problem caused a hang in SFTP transfers from a 287 MessageWay server. 288 289Daniel Stenberg (28 Mar 2010) 290- Ben Greear: If you pass a URL to pop3 that does not contain a message ID as 291 part of the URL, it would previously ask for 'INBOX' which just causes the 292 pop3 server to return an error. 293 294 Now libcurl treats en empty message ID as a request for LIST (list of pop3 295 message IDs). User's code could then parse this and download individual 296 messages as desired. 297 298Daniel Stenberg (27 Mar 2010) 299- Ben Greear brought a patch that from now on allows all protocols to specify 300 name and user within the URL, in the same manner HTTP and FTP have been 301 allowed to in the past - although far from all of the libcurl supported 302 protocls actually have that feature in their URL definition spec. 303 304Daniel Stenberg (26 Mar 2010) 305- Ben Greear brought code that makes the rate limiting code for the easy 306 interface a bit smoother as it introduces sub-second sleeps during it and it 307 also takes the buffer sizes into account. 308 309Daniel Stenberg (24 Mar 2010) 310- Bob Richmond: There's an annoying situation where libcurl will read new HTTP 311 response data from a socket, then check if it's a timeout if one is set. If 312 the last packet received constitutes the end of the response body, libcurl 313 still treats it as a timeout condition and reports a message like: 314 315 "Operation timed out after 3000 milliseconds with 876 out of 876 bytes 316 received" 317 318 It should only a timeout if the timer lapsed and we DIDN'T receive the end 319 of the response body yet. 320 321- Christopher Conroy fixed a problem with RTSP and GET_PARAMETER reported 322 to us by Massimo Callegari. There's a new test case 572 that verifies this 323 now. 324 325- The 'ares' subtree has been removed from the source repository. It was 326 always a separate project that sort of piggybacked on the curl project since 327 the dawn of times and now the time has come for it to go stand on its own 328 legs and continue living its own life. All details on c-ares and its new 329 source code repository is found at http://c-ares.haxx.se/ 330 331Daniel Stenberg (23 Mar 2010) 332- Kenny To filed the bug report #2963679 with patch to fix a problem he 333 experienced with doing multi interface HTTP POST over a proxy using 334 PROXYTUNNEL. He found a case where it would connect fine but bits.tcpconnect 335 was not set correct so libcurl didn't work properly. 336 337 (http://curl.haxx.se/bug/view.cgi?id=2963679) 338 339- Akos Pasztory filed debian bug report #572276 340 http://bugs.debian.org/cgi-bin/bugreport.cgi?bug=572276 mentioning a problem 341 with a resource that returns chunked-encoded _and_ with a Content-Length 342 and libcurl failed to properly ignore the latter information. 343 344- Hauke Duden provided an example program that made the multi interface crash. 345 His example simply used the multi interface and did first one FTP transfer 346 and after completion it used a second easy handle and did another FTP 347 transfer on the same FTP server. 348 349 This triggered a bug in the "delayed easy handle kill" system that curl 350 uses: when an FTP connection is left alive it must keep an easy handle 351 around internally - only for the purpose of having an easy handle when it 352 later disconnects it. The code assumed that when the easy handle was removed 353 and an internal reference was made, that version could be killed later on 354 when a new easy handle came using the same connection. This was wrong as 355 Hauke's example showed that the removed handle wasn't killed for real until 356 later. This caused a double close attempt => segfault. 357 358Daniel Stenberg (22 Mar 2010) 359- Thomas Lopatic fixed the alarm()-based DNS timeout: 360 361 Looking at the code of Curl_resolv_timeout() in hostip.c, I think that in 362 case of a timeout, the signal handler for SIGALRM never gets removed. I 363 think that in my case it gets executed at some point later on when execution 364 has long left Curl_resolv_timeout() or even the cURL library. 365 366 The code that is jumped to with siglongjmp() simply sets the error message 367 to "name lookup timed out" and then returns with CURLRESOLV_ERROR. I guess 368 that instead of simply returning without cleaning up, the code should have a 369 goto that jumps to the spot right after the call to Curl_resolv(). 370 371Kamil Dudka (22 Mar 2010) 372- Douglas Steinwand contributed a patch fixing insufficient initialization in 373 Curl_clone_ssl_config() 374 375Daniel Stenberg (21 Mar 2010) 376- Ben Greear improved TFTP: the error code returning and the treatment 377 of TSIZE == 0 when uploading. 378 379- We've switched from CVS to git. See http://curl.haxx.se/source.html 380 381Kamil Dudka (19 Mar 2010) 382- Improved Curl_read() to not ignore the error returned from Curl_ssl_recv(). 383 384Daniel Stenberg (15 Mar 2010) 385- Constantine Sapuntzakis brought a patch: 386 387 The problem mentioned on Dec 10 2009 388 (http://curl.haxx.se/bug/view.cgi?id=2905220) was only partially fixed. 389 Partially because an easy handle can be associated with many connections in 390 the cache (e.g. if there is a redirect during the lifetime of the easy 391 handle). The previous patch only cleaned up the first one. The new fix now 392 removes the easy handle from all connections, not just the first one. 393 394Daniel Stenberg (6 Mar 2010) 395- Ben Greear brought a patch that fixed the rate limiting logic for TFTP when 396 the easy interface was used. 397 398Daniel Stenberg (5 Mar 2010) 399- Daniel Johnson provided fixes for building curl with the clang compiler. 400 401Yang Tse (5 Mar 2010) 402- Constantine Sapuntzakis detected and fixed a double free in builds done 403 with threaded resolver enabled (Windows default configuration) that would 404 get triggered when a curl handle is closed while doing DNS resolution. 405 406Daniel Stenberg (2 Mar 2010) 407- [Daniel Johnson] I've been trying to build libcurl with clang on Darwin and 408 ran into some issues with the GSSAPI tests in configure.ac. The tests first 409 try to determine the include dirs and libs and set CPPFLAGS and LIBS 410 accordingly. It then checks for the headers and finally sets LIBS a second 411 time, causing the libs to be included twice. The first setting of LIBS seems 412 redundant and should be left out, since the first part is otherwise just 413 about finding headers. 414 415 My second issue is that 'krb5-config --libs gssapi' on Darwin is less than 416 useless and returns junk that, while it happens to work with gcc, causes 417 clang to choke. For example, --libs returns $CFLAGS along with the libs, 418 which is really retarded. Simply setting 'LIBS="$LIBS -lgssapi_krb5 419 -lresolv"' on Darwin is sufficient. 420 421- Based on patch provided by Jacob Moshenko, the transfer logic now properly 422 makes sure that when using sub-second timeouts, there's no final bad 1000ms 423 wait. Previously, a sub-second timeout would often make the elapsed time end 424 up the time rounded up to the nearest second (e.g. 1s for 200ms timeout) 425 426- Andrei Benea filed bug report #2956698 and pointed out that the 427 CURLOPT_CERTINFO feature leaked memory due to a missing OpenSSL function 428 call. He provided the patch to fix it too. 429 430 http://curl.haxx.se/bug/view.cgi?id=2956698 431 432- Markus Duft pointed out in bug #2961796 that even though Interix has a 433 poll() function it doesn't quite work the way we want it so we must disable 434 it, and he also provided a patch for it. 435 436 http://curl.haxx.se/bug/view.cgi?id=2961796 437 438- Made the pingpong timeout code properly deal with the response timeout AND 439 the global timeout if set. Also, as was reported in the bug report #2956437 440 by Ryan Chan, the time stamp to use as basis for the per command timeout was 441 not set properly in the DONE phase for FTP (and not for SMTP) so I fixed 442 that just now. This was a regression compared to 7.19.7 due to the 443 conversion of FTP code over to the generic pingpong concepts. 444 445 http://curl.haxx.se/bug/view.cgi?id=2956437 446 447Daniel Stenberg (1 Mar 2010) 448- Ben Greear provided an update for TFTP that fixes upload. 449 450- Wesley Miaw reported bug #2958179 which identified a case of looping during 451 OpenSSL based SSL handshaking even though the multi interface was used and 452 there was no good reason for it. 453 454 http://curl.haxx.se/bug/view.cgi?id=2958179 455 456Daniel Stenberg (26 Feb 2010) 457- Pat Ray in bug #2958474 pointed out an off-by-one case when receiving a 458 chunked-encoding trailer. 459 460 http://curl.haxx.se/bug/view.cgi?id=2958474 461 462Daniel Fandrich (25 Feb 2010) 463- Fixed a couple of out of memory leaks and a segfault in the SMTP & IMAP code. 464 465Yang Tse (25 Feb 2010) 466- I fixed bug report #2958074 indicating 467 (http://curl.haxx.se/bug/view.cgi?id=2958074) that curl on Windows with 468 option --trace-time did not use local time when timestamping trace lines. 469 This could also happen on other systems depending on time souurce. 470 471Patrick Monnerat (22 Feb 2010) 472- Proper handling of STARTTLS on SMTP, taking CURLUSESSL_TRY into account. 473- SMTP falls back to RFC821 HELO when EHLO fails (and SSL is not required). 474- Use of true local host name (i.e.: via gethostname()) when available, as 475 default argument to SMTP HELO/EHLO. 476- Test case 804 for HELO fallback. 477 478Daniel Stenberg (20 Feb 2010) 479- Fixed the SMTP compliance by making sure RCPT TO addresses are specified 480 properly in angle brackets. Recipients provided with CURLOPT_MAIL_RCPT now 481 get angle bracket wrapping automatically by libcurl unless the recipient 482 starts with an angle bracket as then the app is assumed to deal with that 483 properly on its own. 484 485- I made the SMTP code expect a 250 response back from the server after the 486 full DATA has been sent, and I modified the test SMTP server to also send 487 that response. As usual, the DONE operation that is made after a completed 488 transfer is still not doable in a non-blocking way so this waiting for 250 489 is unfortunately made blockingly. 490 491Yang Tse (14 Feb 2010) 492- Overhauled test suite getpart() function. Fixing potential out of bounds 493 stack and memory overwrites triggered with huge test case definitions. 494 495Daniel Stenberg (13 Feb 2010) 496- Martin Hager reported and fixed a problem with a missing quote in libcurl.m4 497 498 (http://curl.haxx.se/bug/view.cgi?id=2951319) 499 500- Tom Donovan fixed the CURL_FORMAT_* defines when building with cmake. 501 502 (http://curl.haxx.se/bug/view.cgi?id=2951269) 503 504Daniel Stenberg (12 Feb 2010) 505- Jack Zhang reported a problem with SMTP: we wrongly used multiple addresses 506 in the same RCPT TO line, when they should be sent in separate single 507 commands. I updated test case 802 to verify this. 508 509- I also fixed a bad use of my_setopt_str() of CURLOPT_MAIL_RCPT in the curl 510 tool which made it try to output it as string for the --libcurl feature 511 which could lead to crashes. 512 513Yang Tse (11 Feb 2010) 514- Steven M. Schweda fixed VMS builder bad behavior when used in a batch job, 515 removed obsolete batch_compile.com and defines.com and updated VMS readme. 516 517Version 7.20.0 (9 February 2010) 518 519Daniel Stenberg (9 Feb 2010) 520- When downloading compressed content over HTTP and the app asked libcurl to 521 automatically uncompress it with the CURLOPT_ENCODING option, libcurl could 522 wrongly provide the callback with more data than the maximum documented 523 amount. An application could thus get tricked into badness if the maximum 524 limit was trusted to be enforced by libcurl itself (as it is documented). 525 526 This is further detailed and explained in the libcurl security advisory 527 20100209 at 528 529 http://curl.haxx.se/docs/adv_20100209.html 530 531Daniel Fandrich (3 Feb 2010) 532- Changed the Watcom makefiles to make them easier to keep in sync with 533 Makefile.inc since that can't be included directly. 534 535Yang Tse (2 Feb 2010) 536- Symbol CURL_FORMAT_OFF_T now obsoleted, will be removed in a future release, 537 symbol will not be available when building with CURL_NO_OLDIES defined. Use 538 of CURL_FORMAT_CURL_OFF_T is preferred since 7.19.0 539 540Daniel Stenberg (1 Feb 2010) 541- Using the multi_socket API, it turns out at times it seemed to "forget" 542 connections (which caused a hang). It turned out to be an existing (7.19.7) 543 bug in libcurl (that's been around for a long time) and it happened like 544 this: 545 546 The app calls curl_multi_add_handle() to add a new easy handle, libcurl will 547 then set it to timeout in 1 millisecond so libcurl will tell the app about 548 it. 549 550 The app's timeout fires off that there's a timeout, the app calls libcurl as 551 we so often document it: 552 553 do { 554 res = curl_multi_socket_action(... TIMEOUT ...); 555 } while(CURLM_CALL_MULTI_PERFORM == res); 556 557 And this is the problem number one: 558 559 When curl_multi_socket_action() is called with no specific handle, but only 560 a timeout-action, it will *only* perform actions within libcurl that are 561 marked to run at this time. In this case, the request would go from INIT to 562 CONNECT and return CURLM_CALL_MULTI_PERFORM. When the app then calls libcurl 563 again, there's no timer set for this handle so it remains in the CONNECT 564 state. The CONNECT state is a transitional state in libcurl so it reports no 565 sockets there, and thus libcurl never tells the app anything more about that 566 easy handle/connection. 567 568 libcurl _does_ set a 1ms timeout for the handle at the end of 569 multi_runsingle() if it returns CURLM_CALL_MULTI_PERFORM, but since the loop 570 is instant the new job is not ready to run at that point (and there's no 571 code that makes libcurl call the app to update the timout for this new 572 timeout). It will simply rely on that some other timeout will trigger later 573 on or that something else will update the timeout callback. This makes the 574 bug fairly hard to repeat. 575 576 The fix made to adress this issue: 577 578 We introduce a loop in lib/multi.c around all calls to multi_runsingle() and 579 simply check for CURLM_CALL_MULTI_PERFORM internally. This has the added 580 benefit that this goes in line with my long-term wishes to get rid of the 581 CURLM_CALL_MULTI_PERFORM all together from the public API. 582 583 The downside of this fix, is that the counter we return in 'running_handles' 584 in several of our public functions then gets a slightly new and possibly 585 confusing behavior during times: 586 587 If an app adds a handle that fails to connect (very quickly) it may just 588 as well never appear as a 'running_handle' with this fix. Previously it 589 would first bump the counter only to get it decreased again at next call. 590 Even I have used that change in handle counter to signal "end of a 591 transfer". The only *good* way to find the end of a individual transfer 592 is calling curl_multi_info_read() to see if it returns one. 593 594 Of course, if the app previously did the looping before it checked the 595 counter, it really shouldn't be any new effect. 596 597Yang Tse (26 Jan 2010) 598- Constantine Sapuntzakis' and Joshua Kwan's work done in the last four months 599 relative to the asynchronous DNS lookups, along with with some integration 600 adjustments I have done are finally committed to CVS. 601 602 Currently these enhancements will benefit builds done using c-ares on any 603 platform as well as Windows builds using the default threaded resolver. 604 605 This release does not make generally available POSIX threaded DNS lookups 606 yet. There is no configure option to enable this feature yet. It is possible 607 to experimantally try this feature running configure with compiler flags that 608 make simultaneous definition of preprocessor symbols USE_THREADS_POSIX and 609 HAVE_PTHREAD_H, as well as whatever reentrancy compiler flags and linker ones 610 are required to link and properly use pthread_* functions on each platform. 611 612Daniel Stenberg (26 Jan 2010) 613- Mike Crowe made libcurl return CURLE_COULDNT_RESOLVE_PROXY when it is the 614 proxy that cannot be resolved when using c-ares. This matches the behaviour 615 when not using c-ares. 616 617Bj�rn Stenberg (23 Jan 2010) 618- Added a new flag: -J/--remote-header-name. This option tells the 619 -O/--remote-name option to use the server-specified Content-Disposition 620 filename instead of extracting a filename from the URL. 621 622Daniel Stenberg (21 Jan 2010) 623- Chris Conroy brought support for RTSP transfers, and with it comes 8(!) new 624 libcurl options for controlling what to get and how to receive posssibly 625 interleaved RTP data. 626 627Daniel Stenberg (20 Jan 2010) 628- As was pointed out on the http-state mailing list, the order of cookies in a 629 HTTP Cookie: header _needs_ to be sorted on the path length in the cases 630 where two cookies using the same name are set more than once using 631 (overlapping) paths. Realizing this, identically named cookies must be 632 sorted correctly. But detecting only identically named cookies and take care 633 of them individually is harder than just to blindly and unconditionally sort 634 all cookies based on their path lengths. All major browsers also already do 635 this, so this makes our behavior one step closer to them in the cookie area. 636 637 Test case 8 was the only one that broke due to this change and I updated it 638 accordingly. 639 640Daniel Stenberg (19 Jan 2010) 641- David McCreedy brought a fix and a new test case (129) to make libcurl work 642 again when downloading files over FTP using ASCII and it turns out that the 643 final size of the file is not the same as the initial size the server 644 reported. This is very common since servers don't take the newline 645 conversions into account. 646 647Kamil Dudka (14 Jan 2010) 648- Suppressed side effect of OpenSSL configure checks, which prevented NSS from 649 being properly detected under certain circumstances. It had been caused by 650 strange behavior of pkg-config when handling PKG_CONFIG_LIBDIR. pkg-config 651 distinguishes among empty and non-existent environment variable in that case. 652 653Daniel Stenberg (12 Jan 2010) 654- Gil Weber reported a peculiar flaw with the multi interface when doing SFTP 655 transfers: curl_multi_fdset() would return -1 and not set and file 656 descriptors several times during a transfer of a single file. It turned out 657 to be due to two different flaws now fixed. Gil's excellent recipe helped me 658 nail this. 659 660Daniel Stenberg (11 Jan 2010) 661- Made sure that the progress callback is repeatedly called at a regular 662 interval even during very slow connects. 663 664- The tests/runtests.pl script now checks to see if the test case that runs is 665 present in the tests/data/Makefile.am and outputs a notice message on the 666 screen if not. Each test file has to be included in that Makefile.am to get 667 included in release archives and forgetting to add files there is a common 668 mistake. This is an attempt to make it harder to forget. 669 670Daniel Stenberg (9 Jan 2010) 671- Johan van Selst found and fixed a OpenSSL session ref count leak: 672 673 ossl_connect_step3() increments an SSL session handle reference counter on 674 each call. When sessions are re-used this reference counter may be 675 incremented many times, but it will be decremented only once when done (by 676 Curl_ossl_session_free()); and the internal OpenSSL data will not be freed 677 if this reference count remains positive. When a session is re-used the 678 reference counter should be corrected by explicitly calling 679 SSL_SESSION_free() after each consecutive SSL_get1_session() to avoid 680 introducing a memory leak. 681 682 (http://curl.haxx.se/bug/view.cgi?id=2926284) 683 684Daniel Stenberg (7 Jan 2010) 685- Make sure the progress callback is called repeatedly even during very slow 686 name resolves when c-ares is used for resolving. 687 688Claes Jakobsson (6 Jan 2010) 689- Julien Chaffraix fixed so that the fragment part in an URL is not sent 690 to the server anymore. 691 692Kamil Dudka (3 Jan 2010) 693- Julien Chaffraix eliminated a duplicated initialization in singlesocket(). 694 695Daniel Stenberg (2 Jan 2010) 696- Make curl support --ssl and --ssl-reqd instead of the previous FTP-specific 697 versions --ftp-ssl and --ftp-ssl-reqd as these options are now used to 698 control SSL/TLS for IMAP, POP3 and SMTP as well in addition to FTP. The old 699 option names are still working but the new ones are the ones listed and 700 documented. 701 702Daniel Stenberg (1 Jan 2010) 703- Ingmar Runge enhanced libcurl's FTP engine to support the PRET command. This 704 command is a special "hack" used by the drftpd server, but even though it is 705 a custom extension I've deemed it fine to add to libcurl since this server 706 seems to survive and people keep using it and want libcurl to support 707 it. The new libcurl option is named CURLOPT_FTP_USE_PRET, and it is also 708 usable from the curl tool with --ftp-pret. Using this option on a server 709 that doesn't support this command will make libcurl fail. 710 711 I added test cases 1107 and 1108 to verify the functionality. 712 713 The PRET command is documented at 714 http://www.drftpd.org/index.php/Distributed_PASV 715 716Yang Tse (30 Dec 2009) 717- Steven M. Schweda improved VMS build system, and Craig A. Berry helped 718 with the patch and testing. 719 720Daniel Stenberg (26 Dec 2009) 721- Renato Botelho and Peter Pentchev brought a patch that makes the libcurl 722 headers work correctly even on FreeBSD systems before v8. 723 724 (http://curl.haxx.se/bug/view.cgi?id=2916915) 725 726Daniel Stenberg (17 Dec 2009) 727- David Byron fixed Curl_ossl_cleanup to actually call ENGINE_cleanup when 728 available. 729 730- Follow-up fix for the proxy fix I did for Jon Nelson's bug. It turned out I 731 was a bit too quick and broke test case 1101 with that change. The order of 732 some of the setups is sensitive. I now changed it slightly again to make 733 sure we do them in this order: 734 735 1 - parse URL and figure out what protocol is used in the URL 736 2 - prepend protocol:// to URL if missing 737 3 - parse name+password off URL, which needs to know what protocol is used 738 (since only some allows for name+password in the URL) 739 4 - figure out if a proxy should be used set by an option 740 5 - if no proxy option, check proxy environment variables 741 6 - run the protocol-specific setup function, which needs to have the proxy 742 already set 743 744Daniel Stenberg (15 Dec 2009) 745- Jon Nelson found a regression that turned out to be a flaw in how libcurl 746 detects and uses proxies based on the environment variables. If the proxy 747 was given as an explicit option it worked, but due to the setup order 748 mistake proxies would not be used fine for a few protocols when picked up 749 from '[protocol]_proxy'. Obviously this broke after 7.19.4. I now also added 750 test case 1106 that verifies this functionality. 751 752 (http://curl.haxx.se/bug/view.cgi?id=2913886) 753 754Daniel Stenberg (12 Dec 2009) 755- IMAP, POP3 and SMTP support and their TLS versions (including IMAPS, POP3S 756 and SMTPS) are now supported. The current state may not yet be solid, but 757 the foundation is in place and the test suite has some initial support for 758 these protocols. Work will now persue to make them nice libcurl citizens 759 until release. 760 761 The work with supporting these new protocols was sponsored by 762 networking4all.com - thanks! 763 764Daniel Stenberg (10 Dec 2009) 765- Siegfried Gyuricsko found out that the curl manual said --retry would retry 766 on FTP errors in the transient 5xx range. Transient FTP errors are in the 767 4xx range. The code itself only tried on 5xx errors that occured _at login_. 768 Now the retry code retries on all FTP transfer failures that ended with a 769 4xx response. 770 771 (http://curl.haxx.se/bug/view.cgi?id=2911279) 772 773- Constantine Sapuntzakis figured out a case which would lead to libcurl 774 accessing alredy freed memory and thus crash when using HTTPS (with 775 OpenSSL), multi interface and the CURLOPT_DEBUGFUNCTION and a certain order 776 of cleaning things up. I fixed it. 777 778 (http://curl.haxx.se/bug/view.cgi?id=2905220) 779 780Daniel Stenberg (7 Dec 2009) 781- Martin Storsjo made libcurl use the Expect: 100-continue header for posts 782 with unknown size. Previously it was only used for posts with a known size 783 larger than 1024 bytes. 784 785Daniel Stenberg (1 Dec 2009) 786- If the Expect: 100-continue header has been set by the application through 787 curl_easy_setopt with CURLOPT_HTTPHEADER, the library should set 788 data->state.expect100header accordingly - the current code (in 7.19.7 at 789 least) doesn't handle this properly. Martin Storsjo provided the fix! 790 791Yang Tse (28 Nov 2009) 792- Added Diffie-Hellman parameters to several test harness certificate files in 793 PEM format. Required by several stunnel versions used by our test harness. 794 795Daniel Stenberg (28 Nov 2009) 796- Markus Koetter provided a polished and updated version of Chad Monroe's TFTP 797 rework patch that now integrates TFTP properly into libcurl so that it can 798 be used non-blocking with the multi interface and more. BLKSIZE also works. 799 800 The --tftp-blksize option was added to allow setting the TFTP BLKSIZE from 801 the command line. 802 803Daniel Stenberg (26 Nov 2009) 804- Extended and fixed the change I did on Dec 11 for the the progress 805 meter/callback during FTP command/response sequences. It turned out it was 806 really lame before and now the progress meter SHOULD get called at least 807 once per second. 808 809Daniel Stenberg (23 Nov 2009) 810- Bjorn Augustsson reported a bug which made curl not report any problems even 811 though it failed to write a very small download to disk (done in a single 812 fwrite call). It turned out to be because fwrite() returned success, but 813 there was insufficient error-checking for the fclose() call which tricked 814 curl to believe things were fine. 815 816Yang Tse (23 Nov 2009) 817- David Byron modified Makefile.dist vc8 and vc9 targets in order to allow 818 finer granularity control when generating src and lib makefiles. 819 820Yang Tse (22 Nov 2009) 821- I modified configure to force removal of the curlbuild.h file included in 822 distribution tarballs for use by non-configure systems. As intended, this 823 would get overwriten when doing in-tree builds. But VPATH builds would end 824 having two curlbuild.h files, one in the source tree and another in the 825 build tree. With the modification I introduced 5 Nov 2009 this could become 826 an issue when running libcurl's test suite. 827 828Daniel Stenberg (20 Nov 2009) 829- Constantine Sapuntzakis identified a write after close, as the sockets were 830 closed by libcurl before the SSL lib were shutdown and they may write to its 831 socket. Detected to at least happen with OpenSSL builds. 832 833- Jad Chamcham pointed out a bug with connection re-use. If a connection had 834 CURLOPT_HTTPPROXYTUNNEL enabled over a proxy, a subsequent request using the 835 same proxy with the tunnel option disabled would still wrongly re-use that 836 previous connection and the outcome would only be badness. 837 838Yang Tse (18 Nov 2009) 839- I modified the memory tracking system to make it intolerant with zero sized 840 malloc(), calloc() and realloc() function calls. 841 842Daniel Stenberg (17 Nov 2009) 843- Constantine Sapuntzakis provided another fix for the DNS cache that could 844 end up with entries that wouldn't time-out: 845 846 1. Set up a first web server that redirects (307) to a http://server:port 847 that's down 848 2. Have curl connect to the first web server using curl multi 849 850 After the curl_easy_cleanup call, there will be curl dns entries hanging 851 around with in_use != 0. 852 853 (http://curl.haxx.se/bug/view.cgi?id=2891591) 854 855- Marc Kleine-Budde fixed: curl saved the LDFLAGS set during configure into 856 its pkg-config file. So -Wl stuff ended up in the .pc file, which is really 857 bad, and breaks if there are multiple -Wl in our LDFLAGS (which are in 858 PTXdist). bug #2893592 (http://curl.haxx.se/bug/view.cgi?id=2893592) 859 860Kamil Dudka (15 Nov 2009) 861- David Byron improved the configure script to use pkg-config to find OpenSSL 862 (and in particular the list of required libraries) even if a path is given 863 as argument to --with-ssl 864 865Yang Tse (15 Nov 2009) 866- I removed enable-thread / disable-thread configure option. These were only 867 placebo options. The library is always built as thread safe as possible on 868 every system. 869 870Claes Jakobsson (14 Nov 2009) 871- curl-config now accepts '--configure' to see what arguments was 872 passed to the configure script when building curl. 873 874Daniel Stenberg (14 Nov 2009) 875- Claes Jakobsson restored the configure functionality to detect NSS when 876 --with-nss is set but not "yes". 877 878 I think we can still improve that to check for pkg-config in that path etc, 879 but at least this patch brings back the same functionality we had before. 880 881- Camille Moncelier added support for the file type SSL_FILETYPE_ENGINE for 882 the client certificate. It also disable the key name test as some engines 883 can select a private key/cert automatically (When there is only one key 884 and/or certificate on the hardware device used by the engine) 885 886Yang Tse (14 Nov 2009) 887- Constantine Sapuntzakis provided the fix that ensures that an SSL connection 888 won't be reused unless protection level for peer and host verification match. 889 890 I refactored how preprocessor symbol _THREAD_SAFE definition is done. 891 892Kamil Dudka (12 Nov 2009) 893- Kevin Baughman provided a fix preventing libcurl-NSS from crash on doubly 894 closed NSPR descriptor. The issue was hard to find, reported several times 895 before and always closed unresolved. More info at the RH bug: 896 https://bugzilla.redhat.com/534176 897 898- libcurl-NSS now tries to reconnect with TLS disabled in case it detects 899 a broken TLS server. However it does not happen if SSL version is selected 900 manually. The approach was originally taken from PSM. Kaspar Brand helped me 901 to complete the patch. Original bug reports: 902 https://bugzilla.redhat.com/525496 903 https://bugzilla.redhat.com/527771 904 905Yang Tse (12 Nov 2009) 906- I modified configure script to make the getaddrinfo function check also 907 verify if the function is thread safe. 908 909Yang Tse (11 Nov 2009) 910- Marco Maggi reported that compilation failed when configured --with-gssapi 911 and GNU GSS installed due to a missing mutual exclusion of header files in 912 the Kerberos 5 code path. He also verified that my patch worked for him. 913 914Daniel Stenberg (11 Nov 2009) 915- Constantine Sapuntzakis posted bug #2891595 916 (http://curl.haxx.se/bug/view.cgi?id=2891595) which identified how an entry 917 in the DNS cache would linger too long if the request that added it was in 918 use that long. He also provided the patch that now makes libcurl capable of 919 still doing a request while the DNS hash entry may get timed out. 920 921- Christian Schmitz noticed that the progress meter/callback was not properly 922 used during the FTP connection phase (after the actual TCP connect), while 923 it of course should be. I also made the speed check get called correctly so 924 that really slow servers will trigger that properly too. 925 926Kamil Dudka (5 Nov 2009) 927- Dropped misleading timeouts in libcurl-NSS and made sure the SSL socket works 928 in non-blocking mode. 929 930Yang Tse (5 Nov 2009) 931- I removed leading 'curl' path on the 'curlbuild.h' include statement in 932 curl.h, adjusting auto-makefiles include path, to enhance portability to 933 OS's without an orthogonal directory tree structure such as OS/400. 934 935Daniel Stenberg (4 Nov 2009) 936- I fixed several problems with the transfer progress meter. It showed the 937 wrong percentage for small files, most notable for <1000 bytes and could 938 easily end up showing more than 100% at the end. It also didn't show any 939 percentage, transfer size or estimated transfer times when transferring 940 less than 100 bytes. 941 942Version 7.19.7 (4 November 2009) 943 944Daniel Stenberg (2 Nov 2009) 945- As reported independent by both Stan van de Burgt and Didier Brisebourg, 946 CURLINFO_SIZE_DOWNLOAD (the -w variable size_download) didn't work when 947 getting data from ldap! 948 949Daniel Stenberg (31 Oct 2009) 950- Gabriel Kuri reported a problem with CURLINFO_CONTENT_LENGTH_DOWNLOAD if the 951 download was 0 bytes, as libcurl would then return the size as unknown (-1) 952 and not 0. I wrote a fix and test case 566 to verify it. 953 954Daniel Stenberg (30 Oct 2009) 955- Liza Alenchery mentioned a problem with re-used SCP connection when a bad 956 auth is used, as it caused a crash. I failed to repeat the issue, but still 957 made a change that now forces the TCP connection used for a freed SCP 958 session to get closed and not be re-used. 959 960- "Tom" posted a bug report that mentioned how libcurl did wrong when doing a 961 POST using a read callback, with Digest authentication and 962 "Transfer-Encoding: chunked" enforced. I would then cause the first request 963 to be wrongly sent and then basically hang until the server closed the 964 connection. I fixed the problem and added test case 565 to verify it. 965 966Daniel Stenberg (25 Oct 2009) 967- Dima Barsky made the curl cookie parser accept cookies even with blank or 968 unparsable expiry dates and then treat them as session cookies - previously 969 libcurl would reject cookies with a date format it couldn't parse. Research 970 shows that the major browser treat such cookies as session cookies. I 971 modified test 8 and 31 to verify this. 972 973Daniel Stenberg (21 Oct 2009) 974- Attempt to use pkg-config for finding out libssh2 installation details 975 during configure. 976 977- A patch in bug report #2883177 (http://curl.haxx.se/bug/view.cgi?id=2883177) 978 by Johan van Selst introduced the --crlfile option to curl, which makes curl 979 tell libcurl about a file with CRL (certificate revocation list) data to 980 read. 981 982Daniel Stenberg (18 Oct 2009) 983- Ray Dassen provided a patch in Debian's bug tracker (bug number #551461) 984 that now makes curl_getdate(3) actually handles RFC 822 formatted dates that 985 use the "single letter military timezones". 986 http://www.rfc-ref.org/RFC-TEXTS/822/chapter5.html has the details. 987 988- Fixed memory leak in the SCP/SFTP code as it never freed the knownhosts 989 data! 990 991- John Dennis filed bug report #2873666 992 (http://curl.haxx.se/bug/view.cgi?id=2873666) which identified a problem 993 which made libcurl loop infinitely when given incorrect credentials when 994 using HTTP GSS negotiate authentication. He also provided a small and simple 995 patch for it. 996 997- Kevin Baughman found a double close() problem with libcurl-NSS, as when 998 libcurl called NSS to close the SSL "session" it also closed the actual 999 socket. 1000 1001Yang Tse (17 Oct 2009) 1002- Bug report #2866724 indicated 1003 (http://curl.haxx.se/bug/view.cgi?id=2866724) that curl on Windows failed 1004 when writing files whose file names originally contained characters which 1005 are not valid for file names on Windows. Dan Fandrich provided an initial 1006 patch and another revised one to fix this issue. 1007 1008Daniel Stenberg (1 Oct 2009) 1009- Tom Mueller correctly reported in bug report #2870221 1010 (http://curl.haxx.se/bug/view.cgi?id=2870221) that libcurl returned an 1011 incorrect return code from the internal trynextip() function which caused 1012 him grief. This is a regression that was introduced in 7.19.1 and I find it 1013 strange it hasn't hit us harder, but I won't persue into figuring out 1014 exactly why. 1015 1016- Constantine Sapuntzakis: The current implementation will always set 1017 SO_SNDBUF to CURL_WRITE_SIZE even if the SO_SNDBUF starts out larger. The 1018 patch doesn't do a setsockopt if SO_SNDBUF is already greater than 1019 CURL_WRITE_SIZE. This should help folks who have set up their computer with 1020 large send buffers. 1021 1022Daniel Stenberg (27 Sep 2009) 1023- I introduced a maximum limit for received HTTP headers. It is controlled by 1024 the define CURL_MAX_HTTP_HEADER which is even exposed in the public header 1025 file to allow for users to fairly easy rebuild libcurl with a modified 1026 limit. The rationale for a fixed limit is that libcurl is realloc()ing a 1027 buffer to be able to put a full header into it, so that it can call the 1028 header callback with the entire header, but that also risk getting it into 1029 trouble if a server by mistake or willingly sends a header that is more or 1030 less without an end. The limit is set to 100K. 1031 1032Daniel Stenberg (26 Sep 2009) 1033- John P. McCaskey posted a bug report that showed how libcurl did wrong when 1034 saving received cookies with no given path, if the path in the request had a 1035 query part. That is means a question mark (?) and characters on the right 1036 side of that. I wrote test case 1105 and fixed this problem. 1037 1038Kamil Dudka (26 Sep 2009) 1039- Implemented a protocol independent way to specify blocking direction, used by 1040 transfer.c for blocking. It is currently used only by SCP and SFTP protocols. 1041 This enhancement resolves an issue with 100% CPU usage during SFTP upload, 1042 reported by Vourhey. 1043 1044Daniel Stenberg (25 Sep 2009) 1045- Chris Mumford filed bug report #2861587 1046 (http://curl.haxx.se/bug/view.cgi?id=2861587) identifying that libcurl used 1047 the OpenSSL function X509_load_crl_file() wrongly and failed if it would 1048 load a CRL file with more than one certificate within. This is now fixed. 1049 1050Daniel Stenberg (16 Sep 2009) 1051- Sven Anders reported that we introduced a cert verfication flaw for OpenSSL- 1052 powered libcurl in 7.19.6. If there was a X509v3 Subject Alternative Name 1053 field in the certficate it had to match and so even if non-DNS and non-IP 1054 entry was present it caused the verification to fail. 1055 1056Daniel Fandrich (15 Sep 2009) 1057- Moved the libssh2 checks after the SSL library checks. This helps when 1058 statically linking since libssh2 needs the SSL library link flags to be 1059 set up already to satisfy its dependencies. This wouldn't be necessary if 1060 the libssh2 configure check was changed to use pkg-config since the 1061 --static flag would add the dependencies automatically. 1062 1063Yang Tse (14 Sep 2009) 1064- Revert Joshua Kwan's patch committed 11 Sep 2009. 1065 1066 Some systems poll function sets POLLHUP in revents without setting 1067 POLLIN, and sets POLLERR without setting POLLIN and POLLOUT. In some 1068 libcurl code execution paths this could trigger busy wait loops with 1069 high CPU usage until a timeout condition aborted the loop. 1070 1071 The reverted patch addressed the above issue for a very specific case, 1072 when awaiting c-ares to resolve. A libcurl-wide fix for Curl_poll now 1073 superceeds this one. 1074 1075Guenter Knauf (11 Sep 2009) 1076- Joshua Kwan provided a patch to pass POLLERR / POLLHUP back to c-ares. 1077 This fixes a loop problem with high CPU usage. 1078 1079Daniel Stenberg (10 Sep 2009) 1080- Claes Jakobsson fixed a problem with cookie expiry dates at exctly the epoch 1081 start second "Thu Jan 1 00:00:00 GMT 1970" as the date parser then returns 0 1082 which internally then is treated as a session cookie. That particular date 1083 is now made to get the value of 1. 1084 1085Daniel Stenberg (2 Sep 2009) 1086- Daniel Johnson found a flaw in the code converting sftp-errors to libcurl 1087 errors. 1088 1089Daniel Stenberg (1 Sep 2009) 1090- Peter Sylvester made a debug feature for Curl_resolv() that now will force 1091 libcurl to resolve 'localhost' whatever name you use in the URL *if* you set 1092 the --interface option to (exactly) "LocalHost". This will enable us to 1093 write tests for custom hosts names but still use a local host server. 1094 1095- configure now tries to use pkg-config for a number of sub-dependencies even 1096 when cross-compiling. The key to success is then you properly setup 1097 PKG_CONFIG_PATH before invoking configure. 1098 1099 I also improved how NSS is detected by trying nss-config if pkg-config isn't 1100 present, and as a last resort just use the lib name and force the user to 1101 setup the LIBS/LDFLAGS/CFLAGS etc properly. The previous last resort would 1102 add a range of various libs that would almost never be quite correct. 1103 1104Daniel Stenberg (31 Aug 2009) 1105- When using the multi interface with FTP and you asked for NOBODY, you did no 1106 QUOTE commands and the request used the same path as the connection had 1107 already changed to, it would decide that no commands would be necessary for 1108 the "DO" action and that was not handled properly but libcurl would instead 1109 hang. 1110 1111Kamil Dudka (28 Aug 2009) 1112- Improved error message for not matching certificate subject name in 1113 libcurl-NSS. Originally reported at: 1114 https://bugzilla.redhat.com/show_bug.cgi?id=516056#c9 1115 1116Patrick Monnerat (24 Aug 2009) 1117- Introduced a SYST-based test to properly set-up name format when dealing 1118 with the OS/400 FTP server. 1119 1120- Fixed an ftp_readresp() bug preventing detection of failing control socket 1121 and causing FTP client to loop forever. 1122 1123Daniel Stenberg (24 Aug 2009) 1124- Marc de Bruin pointed out that configure --with-gnutls=PATH didn't work 1125 properly and provided a fix. http://curl.haxx.se/bug/view.cgi?id=2843008 1126 1127- Eric Wong introduced support for the new option -T. (dot) that makes curl 1128 read stdin in a non-blocking fashion. This also brings back -T- (minus) to 1129 the previous blocking behavior since it could break stuff for people at 1130 times. 1131 1132Michal Marek (21 Aug 2009) 1133- With CURLOPT_PROXY_TRANSFER_MODE, avoid sending invalid URLs like 1134 ftp://example.com;type=i if the user specified ftp://example.com without the 1135 slash. 1136 1137Daniel Stenberg (21 Aug 2009) 1138- Andre Guibert de Bruet pointed out a missing return code check for a 1139 strdup() that could lead to segfault if it returned NULL. I extended his 1140 suggest patch to now have Curl_retry_request() return a regular return code 1141 and better check that. 1142 1143- Lots of good work by Krister Johansen, mostly related to pipelining: 1144 1145 Fix SIGSEGV on free'd easy_conn when pipe unexpectedly breaks 1146 Fix data corruption issue with re-connected transfers 1147 Fix use after free if we're completed but easy_conn not NULL 1148 1149Kamil Dudka (13 Aug 2009) 1150- Changed NSS code to not ignore the value of ssl.verifyhost and produce more 1151 verbose error messages. Originally reported at: 1152 https://bugzilla.redhat.com/show_bug.cgi?id=516056 1153 1154Daniel Stenberg (12 Aug 2009) 1155- Karl Moerder fixed the Makefile.vc* makefiles to include the new file 1156 nonblock.c so that they work fine again 1157 1158- I expanded test 517 with a bunch of more dates that originate from the 1159 Chrome browser test suite. It turns out most of them get parsed the same 1160 way. 1161 1162Version 7.19.6 (12 August 2009) 1163 1164Daniel Stenberg (12 Aug 2009) 1165- Carsten Lange reported a bug and provided a patch for TFTP upload and the 1166 sending of the TSIZE option. I don't like fixing bugs just hours before 1167 a release, but since it was broken and the patch fixes this for him I decided 1168 to get it in anyway. 1169 1170Daniel Stenberg (11 Aug 2009) 1171- Peter Sylvester made the HTTPS test server use specific certificates for 1172 each test, so that the test suite can now be used to actually test the 1173 verification of cert names etc. This made an error show up in the OpenSSL- 1174 specific code where it would attempt to match the CN field even if a 1175 subjectAltName exists that doesn't match. This is now fixed and verified 1176 in test 311. 1177 1178- Benbuck Nason posted the bug report #2835196 1179 (http://curl.haxx.se/bug/view.cgi?id=2835196), fixing a few compiler 1180 warnings when mixing ints and bools. 1181 1182Daniel Fandrich (10 Aug 2009) 1183- Fixed a memory leak in the FTP code and an off-by-one heap buffer overflow. 1184 1185Daniel Fandrich (9 Aug 2009) 1186- Fixed some memory leaks in the command-line tool that caused most of the 1187 torture tests to fail. 1188 1189Daniel Stenberg (2 Aug 2009) 1190- Curt Bogmine reported a problem with SNI enabled on a particular server. We 1191 should introduce an option to disable SNI, but as we're in feature freeze 1192 now I've addressed the obvious bug here (pointed out by Peter Sylvester): we 1193 shouldn't try to enable SNI when SSLv2 or SSLv3 is explicitly selected. 1194 Code for OpenSSL and GnuTLS was fixed. NSS doesn't seem to have a particular 1195 option for SNI, or are we simply not using it? 1196 1197Daniel Stenberg (1 Aug 2009) 1198- Scott Cantor posted the bug report #2829955 1199 (http://curl.haxx.se/bug/view.cgi?id=2829955) mentioning the recent SSL cert 1200 verification flaw found and exploited by Moxie Marlinspike. The presentation 1201 he did at Black Hat is available here: 1202 https://www.blackhat.com/html/bh-usa-09/bh-usa-09-archives.html#Marlinspike 1203 1204 Apparently at least one CA allowed a subjectAltName or CN that contain a 1205 zero byte, and thus clients that assumed they would never have zero bytes 1206 were exploited to OK a certificate that didn't actually match the site. Like 1207 if the name in the cert was "example.com\0theatualsite.com", libcurl would 1208 happily verify that cert for example.com. 1209 1210 libcurl now better uses the length of the extracted name, not using the zero 1211 termination for getting the string length. 1212 1213 This fixing only made and needed in OpenSSL interfacing code. 1214 1215- Tanguy Fautre pointed out that OpenSSL's function RAND_screen() (present 1216 only in some OpenSSL installs - like on Windows) isn't thread-safe and we 1217 agreed that moving it to the global_init() function is a decent way to deal 1218 with this situation. 1219 1220- Alexander Beedie provided the patch for a noproxy problem: If I have set 1221 CURLOPT_NOPROXY to "*", or to a host that should not use a proxy, I actually 1222 could still end up using a proxy if a proxy environment variable was set. 1223 1224Daniel Stenberg (27 Jul 2009) 1225- All the quote options (CURLOPT_QUOTE, CURLOPT_POSTQUOTE and 1226 CURLOPT_PREQUOTE) now accept a preceeding asterisk before the command to 1227 send when using FTP, as a sign that libcurl shall simply ignore the response 1228 from the server instead of treating it as an error. Not treating a 400+ FTP 1229 response code as an error means that failed commands will not abort the 1230 chain of commands, nor will they cause the connection to get disconnected. 1231 1232Daniel Stenberg (26 Jul 2009) 1233- Johan van Selst posted bug report #2825989 1234 (http://curl.haxx.se/bug/view.cgi?id=2825989) pointing out that 1235 OpenSSL-powered libcurl didn't support the SHA-2 digest algorithm, and 1236 provided the solution too: to use OpenSSL_add_all_algorithms() in addition 1237 to the older SSLeay_* alternative. OpenSSL_add_all_algorithms was added in 1238 OpenSSL 0.9.5 1239 1240Daniel Stenberg (23 Jul 2009) 1241- Added CURLOPT_SSH_KNOWNHOSTS, CURLOPT_SSH_KEYFUNCTION, CURLOPT_SSH_KEYDATA. 1242 They introduce known_host support for SSH keys to libcurl. See docs for 1243 details. Note that this feature depends on a new enough libssh2 version, to 1244 be supported in libssh2 1.2 and later (or current git repo at this time). 1245 1246Michal Marek (22 Jul 2009) 1247- David Binderman found a memory and fd leak in lib/gtls.c:load_file() 1248 (https://bugzilla.novell.com/523919). When looking at the code, I found that 1249 also the ptr pointer can leak. 1250 1251Kamil Dudka (20 Jul 2009) 1252- Claes Jakobsson improved the support for client certificates handling in 1253 NSS-powered libcurl. Now the client certificates can be selected 1254 automatically by a NSS built-in hook. Additionally pre-login to all PKCS11 1255 slots is no more performed. It used to cause problems with HW tokens. 1256 1257- Fixed reference counting for NSS client certificates. Now the PEM reader 1258 module should be always properly unloaded on Curl_nss_cleanup(). If the 1259 unload fails though, libcurl will try to reuse the already loaded instance. 1260 1261Daniel Fandrich (15 Jul 2009) 1262- Added nonblock.c to the non-automake makefiles (note that the dependencies 1263 in the Watcom makefiles aren't quite correct). 1264 1265Michal Marek (15 Jul 2009) 1266- Changed the description of CURLINFO_OS_ERRNO to make it clear that the 1267 errno is not reset on success. 1268 1269Guenter Knauf (14 Jul 2009) 1270- renamed generated config.h to curl_config.h to avoid any future clashes 1271 with config.h from other projects. 1272 1273Daniel Stenberg (9 Jul 2009) 1274- Eric Wong introduced curlx_nonblock() that the curl tool now (re-)uses for 1275 setting a file descriptor non-blocking. Used by the functionality Eric 1276 himself brough on June 15th. 1277 1278Daniel Stenberg (8 Jul 2009) 1279- Constantine Sapuntzakis posted bug report #2813123 1280 (http://curl.haxx.se/bug/view.cgi?id=2813123) and an a patch that fixes the 1281 problem: 1282 1283 Url A is accessed using auth. Url A redirects to Url B (on a different 1284 server0. Url B reuses a persistent connection. Url B has auth, even though 1285 it's on a different server. 1286 1287 Note: if Url B does not reuse a persistent connection, auth is not sent. 1288 1289 reason: 1290 1291 data->state.first_host is not initialized becuase Curl_http_connect is not 1292 called when a connection is reused. 1293 1294 Solution: 1295 1296 move initialization of data->state.first_host to Curl_http. No code before 1297 Curl_http uses data->state.first_host anyway. 1298 1299Guenter Knauf (4 Jul 2009) 1300- Markus Koetter provided a patch to avoid getnameinfo() usage which broke a 1301 couple of both IPv4 and IPv6 autobuilds. 1302 1303Daniel Stenberg (29 Jun 2009) 1304- Markus Koetter made CURLOPT_FTPPORT (and curl's -P/--ftpport) support a port 1305 range if given colon-separated after the host name/address part. Like 1306 "192.168.0.1:2000-10000" 1307 1308- Modified the separators used for CURLOPT_CERTINFO in multi-part outputs. I 1309 don't know how they got wrong in the first place, but using this output 1310 format makes it possible to quite easily separate the string into an array 1311 of multiple items. 1312 1313Daniel Fandrich (16 June 2009) 1314- Added a few more compiler warning options for gcc. 1315 1316Daniel Stenberg (16 Jun 2009) 1317- Reuven Wachtfogel made curl -o - properly produce a binary output on windows 1318 (no newline translations). Use -B/--use-ascii if you rather get the ascii 1319 approach. 1320 1321Michal Marek (16 Jun 2009) 1322- When doing non-anonymous ftp via http proxies and the password is not 1323 provided in the url, add it there (squid needs this). 1324 1325Daniel Stenberg (15 Jun 2009) 1326- Eric Wong's patch: 1327 1328 This allows curl(1) to be used as a client-side tunnel for arbitrary stream 1329 protocols by abusing chunked transfer encoding in both the HTTP request and 1330 HTTP response. This requires server support for sending a response while a 1331 request is still being read, of course. 1332 1333 If attempting to read from stdin returns EAGAIN, then we pause our sender. 1334 This leaves curl to attempt to read from the socket while reading from stdin 1335 (and thus sending) is paused. 1336 1337 This change was needed to allow successfully tunneling the git protocol over 1338 HTTP (--no-buffer is needed, as well). 1339 1340Patrick Monnerat (15 Jun 2009) 1341- Replaced use of standard C library rand()/srand() by our own pseudo-random 1342 number generator. 1343 1344Yang Tse (11 Jun 2009) 1345- I adapted testcurl script to allow building test harness programs when 1346 cross-compiling for a *-*-mingw* host. 1347 1348Daniel Stenberg (10 Jun 2009) 1349- Fabian Keil ran clang on the (lib)curl code, found a bunch of warnings and 1350 contributed a range of patches to fix them. 1351 1352Yang Tse (10 Jun 2009) 1353- I introduced configure script option --enable-curldebug which now allows 1354 the decoupled enabling or disabling of the curl debug memory tracking 1355 feature from the --enable-debug option which no longer controls this. 1356 1357 curl --version will list 'Debug' feature for debug enabled builds, and 1358 will list 'TrackMemory' feature for curl debug memory tracking capable 1359 builds. These features are independent and can be controlled when running 1360 the configure script. When --enable-debug is given both features will be 1361 enabled, unless some restriction prevents memory tracking from being used. 1362 1363 Internally, definition of preprocessor symbol DEBUGBUILD restricts code 1364 which is only compiled for debug enabled builds. And symbol CURLDEBUG is 1365 used to differentiate code which is _only_ used for memory tracking. 1366 1367Yang Tse (9 Jun 2009) 1368- Daniel Steinberg pointed out that Curl_FormInit() in formdata.c was not 1369 initializing the fread callback pointer and this triggered a compiler 1370 warning, also provided a friendly suggestion on how to fix it. 1371 1372Daniel Stenberg (8 Jun 2009) 1373- Claes Jakobsson provided a patch for libcurl-NSS that fixed a bad refcount 1374 issue with client certs that caused issues like segfaults. 1375 http://curl.haxx.se/mail/lib-2009-05/0316.html 1376 1377- Triggered by bug report #2798852 and the patch in there, I fixed configure 1378 to detect gnutls build options with pkg-config only and not libgnutls-config 1379 anymore since GnuTLS has stopped distributing that tool. If an explicit path 1380 is given to configure, we will instead guess on how to link and use that 1381 lib. I did not use the patch from the bug report. 1382 1383Yang Tse (8 Jun 2009) 1384- Igor Novoseltsev adjusted Makefile.vxworks to get sources and headers 1385 included from Makefile.inc, and provided docs\INSTALL VxWorks section. 1386 1387- I removed buildconf.bat from release and daily snapshot archives. This 1388 file is only for CVS tree checkout builds. 1389 1390Daniel Stenberg (8 Jun 2009) 1391- Eric Wong fixed --no-buffer to actually switch off output buffering. Been 1392 broken since 7.19.0 1393 1394Bill Hoffman (6 Jun 2009) 1395- Added some cmake docs and fixed socklen_t in the build. 1396 1397Yang Tse (5 Jun 2009) 1398- John E. Malmberg provided VMS specific patch: "This fixes an existing bug 1399 in urlglob.c where it was not converting the Curl Unix exit code to a VMS 1400 DCL compatible exit code. This fix required the enhancement described next. 1401 This also adds an enhancement to main.c so that when curl is run under a 1402 Unix shell like Bash on VMS, it will return the standard Unix exit codes 1403 and messages." And another patch for docs/examples. 1404 1405 I introduced os-specific.c and os-specific.h for use in curl tool code 1406 and adjusted John E. Malmberg's patch placement to use these new files 1407 as an effort to prevent main.c from growing ad infinitum. Code already 1408 existing in main.c which is OS specific should be moved into these files. 1409 1410Daniel Stenberg (4 June 2009) 1411- Setting the Content-Length: header from your app when you do a POST or PUT 1412 is almost always a VERY BAD IDEA. Yet there are still apps out there doing 1413 this, and now recently it triggered a bug/side-effect in libcurl as when 1414 libcurl sends a POST or PUT with NTLM, it sends an empty post first when it 1415 knows it will just get a 401/407 back. If the app then replaced the 1416 Content-Length header, it caused the server to wait for input that libcurl 1417 wouldn't send. Aaron Oneal reported this problem in bug report #2799008 1418 (http://curl.haxx.se/bug/view.cgi?id=2799008) and helped us verify the fix. 1419 1420Yang Tse (4 Jun 2009) 1421- Igor Novoseltsev provided patches and information, that after some 1422 adjustments to better fit curl's way of doing things, have resulted 1423 in the posibility of building libcurl for VxWorks. 1424 1425Daniel Fandrich (2 June 2009) 1426- Checked in a Google Android make file. To use it, you must first 1427 create a config.h file by running configure in the Android environment, 1428 which doesn't seem to be easy to do. If no easy way can be found, a 1429 static config-android.h may need to be created and checked in to the 1430 libcurl source tree. 1431 1432Daniel Stenberg (1 June 2009) 1433- Claes Jakobsson fixed the configure script to better find and use NSS 1434 without pkg-config. 1435 1436Yang Tse (1 Jun 2009) 1437- John E. Malmberg provided a VMS specific clean-up for curl.h, and pointed 1438 out that the configure script was failing to detect the timeval struct on 1439 VMS when building with _XOPEN_SOURCE_EXTENDED undefined due to definition 1440 taking place in socket.h instead of time.h. I have adjusted configure 1441 script to also include this header when checking struct timeval. 1442 1443Daniel Stenberg (27 May 2009) 1444- Frank McGeough provided a small OpenSSL #include fix to make libcurl compile 1445 fine with Nokia 5th edition 1.0 SDK for Symbian. 1446 1447- Andre Guibert de Bruet found a call to a OpenSSL function that didn't check 1448 for a failure properly. 1449 1450- Mike Crowe pointed out that setting CURLOPT_USERPWD to NULL used to clear 1451 the auth credentials back in 7.19.0 and earlier while now you have to set "" 1452 to get the same effect. His patch brings back the ability to use NULL. 1453 1454- Claes Jakobsson fixed libcurl-NSS to build fine even without the 1455 PK11_CreateGenericObject() function. 1456 1457Daniel Stenberg (25 May 2009) 1458- bug report #2796358 (http://curl.haxx.se/bug/view.cgi?id=2796358) pointed 1459 out that the cookie parser would leak memory when it parses cookies that are 1460 received with domain, path etc set multiple times in the same header. While 1461 such a cookie is questionable, they occur in the wild and libcurl no longer 1462 leaks memory for them. I added such a header to test case 8. 1463 1464Daniel Fandrich (22 May 2009) 1465- Removed some obsolete digest code that caused a valgrind error in test 551. 1466 1467Daniel Fandrich (20 May 2009) 1468- Added "non-existing host" test keywords to make it easy to skip those 1469 tests on machines that have broken DNS configurations (such as 1470 those configured to use OpenDNS). 1471 1472Daniel Stenberg (19 May 2009) 1473- Kamil Dudka brought the patch from the Redhat bug entry 1474 https://bugzilla.redhat.com/show_bug.cgi?id=427966 which was libcurl closing 1475 a bad file descriptor when closing down the FTP data connection. Caolan 1476 McNamara seems to be the original author of it. 1477 1478Version 7.19.5 (18 May 2009) 1479 1480Daniel Stenberg (17 May 2009) 1481- James Bursa posted a patch to the mailing list that fixed a problem with 1482 no_proxy which made it not skip the proxy if the URL entered contained a 1483 user name. I added test case 1101 to verify. 1484 1485Daniel Stenberg (11 May 2009) 1486- Balint Szilakszi reported a memory leak when libcurl did gzip decompression 1487 of streams that had some parts (legitimately) missing. We now provide and use 1488 a proper cleanup function for the content encoding submodule. 1489 http://curl.haxx.se/mail/lib-2009-05/0092.html 1490 1491- Kamil Dudka provided a fix for libcurl-NSS reported by Michael Cronenworth 1492 at https://bugzilla.redhat.com/show_bug.cgi?id=453612#c12 1493 1494 If an incorrect password is given while loading a private key, libcurl ends 1495 up in an infinite loop consuming memory. The bug is critical. 1496 1497- I fixed the problem with doing NTLM, POST and then following a 302 redirect, 1498 as reported by Ebenezer Ikonne (on curl-users) and Laurent Rabret (on 1499 curl-library). The transfer was mistakenly marked to get more data to send 1500 but since it didn't actually have that, it just hung there... 1501 1502Daniel Stenberg (10 May 2009) 1503- Andre Guibert de Bruet correctly pointed out an over-alloc with one wasted 1504 byte in the digest code. 1505 1506Yang Tse (9 May 2009) 1507- Removed DOS and TPF package's subdirectory Makefile.am, it was only used 1508 to include some files in the distribution tarball serving no other purpose. 1509 Files from the DOS and TPF subdirectories are now included in the EXTRA_DIST 1510 of the Makefile in the parent subdirectory. 1511 1512Yang Tse (8 May 2009) 1513- Changed host name literal in several tests to one under the haxx.se domain. 1514 1515- Renamed vc6 workspace and project files to avoid filename clash when used 1516 for conversion to later VS versions. 1517 1518Daniel Stenberg (8 May 2009) 1519- Constantine Sapuntzakis fixed bug report #2784055 1520 (http://curl.haxx.se/bug/view.cgi?id=2784055) identifying a problem to 1521 connect to SOCKS proxies when using the multi interface. It turned out to 1522 almost not work at all previously. We need to wait for the TCP connect to 1523 be properly verified before doing the SOCKS magic. 1524 1525 There's still a flaw in the FTP code for this. 1526 1527Daniel Stenberg (7 May 2009) 1528- Made the SO_SNDBUF setting for the data connection socket for ftp uploads as 1529 well. See change 28 Apr 2009. 1530 1531Yang Tse (7 May 2009) 1532- Fixed an issue affecting FTP transfers, introduced with the transfer.c 1533 patch committed May 4. 1534 1535Daniel Stenberg (7 May 2009) 1536- Man page *roff problems fixed thanks to input from Colin Watson. Problems 1537 reported in the Debian package. 1538 1539- Vijay G filed bug report #2723236 1540 (http://curl.haxx.se/bug/view.cgi?id=2723236) identifying a problem with 1541 libcurl's TFTP code and its lack of dealing with the OACK packet. 1542 1543Yang Tse (5 May 2009) 1544- Fixed the --ftp-port address of test #251 to the CLIENTIP address, and 1545 reverted the change affecting test suite harness committed 4 May. 1546 1547Daniel Stenberg (5 May 2009) 1548- Inspired by Michael Smith's session id fix for OpenSSL, I did the 1549 corresponding fix in the GnuTLS code: make sure to store the new session id 1550 in case the previous re-used one is rejected. 1551 1552Daniel Stenberg (4 May 2009) 1553- Michael Smith posted bug report #2786255 1554 (http://curl.haxx.se/bug/view.cgi?id=2786255) with a patch, identifying how 1555 libcurl did not deal with SSL session ids properly if the server rejected a 1556 re-use of one. Starting now, it will forget the rejected one and remember 1557 the new. This change was for OpenSSL only, it is likely that other SSL lib 1558 code needs similar fixes. 1559 1560Yang Tse (4 May 2009) 1561- Applied David McCreedy's "transfer.c fixes for CURL_DO_LINEEND_CONV and 1562 non-ASCII platform HTTP requests" patch addressing two HTTP PUT problems: 1563 1) On non-ASCII platforms not all of the protocol portions of the PUT are 1564 being translated to ASCII. 2) On all platforms the line endings of part of 1565 the protocol portions are mangled from CRLF to CRCRLF if data->set.crlf or 1566 data->set.prefer_ascii are set (depending on CURL_DO_LINEEND_CONV). 1567 1568- Applied David McCreedy's patch to fix test suite harness to allow test FTP 1569 server and client on different machines, providing FTP client address when 1570 running the FTP test server. 1571 1572Daniel Fandrich (3 May 2009) 1573- Added and disabled test case 563 which shows KNOWN_BUGS #59. The bug 1574 report failed to mention that a proxy must be used to reproduce it. 1575 1576Yang Tse (2 May 2009) 1577- Use a build-time configured curl_socklen_t data type instead of socklen_t. 1578 1579Yang Tse (1 May 2009) 1580- Applied David McCreedy's patches "TPF-platform specific changes to various 1581 files" and "http.c fix to Curl_proxyCONNECT for non-ASCII platforms", the 1582 former with minor edits. 1583 1584Daniel Stenberg (30 Apr 2009) 1585- I was going to fix issue #59 in KNOWN_BUGS 1586 1587 If the CURLOPT_PORT option is used on an FTP URL like 1588 "ftp://example.com/file;type=A" the ";type=A" is stripped off. 1589 1590 I added test case 562 to verify, only to find out that I couldn't repeat 1591 this bug so I hereby consider it not a bug anymore! 1592 1593Daniel Stenberg (29 Apr 2009) 1594- Based on bug report #2723219 (http://curl.haxx.se/bug/view.cgi?id=2723219) 1595 I've now made TFTP "connections" not being kept for re-use within libcurl. 1596 TFTP is UDP-based so the benefit was really low (if even existing) to begin 1597 with so instead of tracking down to fix this problem we instead removed the 1598 re-use. I also enabled test case 1099 that I wrote a few days ago to verify 1599 that this change fixes the reported problem. 1600 1601Daniel Stenberg (28 Apr 2009) 1602- Constantine Sapuntzakis filed bug report #2783090 1603 (http://curl.haxx.se/bug/view.cgi?id=2783090) pointing out that on windows 1604 we need to grow the SO_SNDBUF buffer somewhat to get really good upload 1605 speeds. http://support.microsoft.com/kb/823764 has the details. Friends 1606 confirmed that simply adding 32 to CURL_MAX_WRITE_SIZE is enough. 1607 1608- Bug report #2709004 (http://curl.haxx.se/bug/view.cgi?id=2709004) by Tim 1609 Chen pointed out how curl couldn't upload with resume when reading from a 1610 pipe. 1611 1612 This ended up with the introduction of a new return code for the 1613 CURLOPT_SEEKFUNCTION callback that basically says that the seek failed but 1614 that libcurl may try to resolve the situation anyway. In our case this means 1615 libcurl will attempt to instead read that much data from the stream instead 1616 of seeking and that way curl can now upload with resume when data is read 1617 from a stream! 1618 1619Daniel Stenberg (26 Apr 2009) 1620- Bug report #2779733 (http://curl.haxx.se/bug/view.cgi?id=2779733) by Sven 1621 Wegener pointed out that CURLINFO_APPCONNECT_TIME didn't work with the multi 1622 interface and provided a patch that fixed the problem! 1623 1624Daniel Stenberg (24 Apr 2009) 1625- Kamil Dudka fixed another NSS-related leak when client certs were used. 1626 1627- Bug report #2779245 (http://curl.haxx.se/bug/view.cgi?id=2779245) by Rainer 1628 Koenig pointed out that the man page didn't tell that the *_proxy 1629 environment variables can be specified lower case or UPPER CASE and the 1630 lower case takes precedence, 1631 1632Daniel Fandrich (21 Apr 2009) 1633- Added new libcurl source files to Amiga, RiscOS and VC6 build files. 1634 1635Yang Tse (21 Apr 2009) 1636- Moved potential inclusion of system's malloc.h and memory.h header files to 1637 setup_once.h. Inclusion of each header file is based on the definition of 1638 NEED_MALLOC_H and NEED_MEMORY_H respectively. 1639 1640 Renamed libcurl's memory.h to curl_memory.h 1641 1642Daniel Stenberg (20 Apr 2009) 1643- Leanic Lefever reported a crash and did some detailed research on why and 1644 how it occurs (http://curl.haxx.se/mail/lib-2009-04/0289.html). The 1645 conclusion was that if an error is detected and Curl_done() is called for 1646 the connection, ftp_done() could at times return another error code that 1647 then would take precedence and that new code confused existing logic that 1648 works for the first error code (CURLE_SEND_ERROR) only. 1649 1650- Gisle Vanem noticed that --libtool would produce bogus strings at times for 1651 OBJECTPOINT options. Now we've introduced a new function - my_setopt_str - 1652 within the app for setting plain string options to avoid the risk of this 1653 mistake happening. 1654 1655Daniel Stenberg (17 Apr 2009) 1656- Pramod Sharma reported and tracked down a bug when doing FTP over a HTTP 1657 proxy. libcurl would then wrongly close the connection after each 1658 request. In his case it had the weird side-effect that it killed NTLM auth 1659 for the proxy causing an inifinite loop! 1660 1661 I added test case 1098 to verify this fix. The test case does however not 1662 properly verify that the transfers are done persistently - as I couldn't 1663 think of a clever way to achieve it right now - but you need to read the 1664 stderr output after a test run to see that it truly did the right thing. 1665 1666Daniel Stenberg (13 Apr 2009) 1667- bug report #2727981 (http://curl.haxx.se/bug/view.cgi?id=2727981) by Martin 1668 Storsj� pointed out how setting CURLOPT_NOBODY to 0 could be downright 1669 confusing as it set the method to either GET or HEAD. The example he showed 1670 looked like: 1671 1672 curl_easy_setopt(curl, CURLOPT_PUT, 1); 1673 curl_easy_setopt(curl, CURLOPT_NOBODY, 0); 1674 1675 The new way doesn't alter the method until the request is about to start. If 1676 CURLOPT_NOBODY is then 1 the HTTP request will be HEAD. If CURLOPT_NOBODY is 1677 0 and the request happens to have been set to HEAD, it will then instead be 1678 set to GET. I believe this will be less surprising to users, and hopefully 1679 not hit any existing users badly. 1680 1681- Toshio Kuratomi reported a memory leak problem with libcurl+NSS that turned 1682 out to be leaking cacerts. Kamil Dudka helped me complete the fix. The issue 1683 is found in Redhat's bug tracker: 1684 https://bugzilla.redhat.com/show_bug.cgi?id=453612 1685 1686 There are still memory leaks present, but they seem to have other reasons. 1687 1688Daniel Fandrich (11 Apr 2009) 1689- Added new libcurl source files to Symbian OS build files. 1690- Improved Symbian support for SSL. 1691 1692Yang Tse (10 Apr 2009) 1693- Daniel Johnson improved the MacOSX-Framework shell script to now perform all 1694 the steps required to build a Mac OS X four way fat ppc/i386/ppc64/x86_64 1695 libcurl.framework. Four way fat framework requires OS X 10.5 SDK or later. 1696 1697Yang Tse (8 Apr 2009) 1698- Removed Sun compilers preprocessor block from curlbuild.h.dist, this also 1699 removes it from the curlbuild.h file originally distributed by the cURL 1700 project as this file is intended for systems not capable of running the 1701 configure script. For those who have been building curl out of the source 1702 code curl distribution tarball provided by curl.haxx.se the change implies 1703 nothing. Previous change in this area committed 2 Apr becomes irrelevant. 1704 1705Daniel Stenberg (6 Apr 2009) 1706- I clarified in the docs that CURLOPT_SEEKFUNCTION should return 0 on success 1707 and 1 on fatal errors. Previously it only mentioned non-zero on fatal 1708 errors. This is a slight change in meaning, but it follows what we've done 1709 elsewhere before and it opens up for LOTS of more useful return codes 1710 whenever we can think of them... 1711 1712Yang Tse (2 Apr 2009) 1713- Fix curl_off_t definition for builds done using Sun compilers and a 1714 non-configured libcurl. In this case curl_off_t data type was gated 1715 to the off_t data type which depends on the _FILE_OFFSET_BITS. This 1716 configuration is exactly the unwanted configuration for our curl_off_t 1717 data type which must not depend on such setting. This breaks ABI for 1718 libcurl libraries built with Sun compilers which were built without 1719 having run the configure script with _FILE_OFFSET_BITS different than 1720 64 and using the ILP32 data model. 1721 1722Daniel Stenberg (1 Apr 2009) 1723- Andre Guibert de Bruet fixed a NULL pointer use in an infof() call if a 1724 strdup() call failed. 1725 1726Daniel Fandrich (31 Mar 2009) 1727- Properly return an error code in curl_easy_recv (reported by Jim Freeman). 1728 1729Daniel Stenberg (18 Mar 2009) 1730- Kamil Dudka brought a patch that enables 6 additional crypto algorithms when 1731 NSS is used. These ciphers were added in NSS 3.4 and require to be enabled 1732 explicitly. 1733 1734Daniel Stenberg (13 Mar 2009) 1735- Use libssh2_version() to present the libssh2 version in case the libssh2 1736 library is found to support it. 1737 1738Yang Tse (12 Mar 2009) 1739- Added missing Curl_read() return code checking in TELNET transfers. 1740 1741- Pierre Brico found and fixed TELNET transfers not being aborted upon 1742 a write callback failure. 1743 1744Daniel Stenberg (11 Mar 2009) 1745- Kamil Dudka made the curl tool properly call curl_global_init() before any 1746 other libcurl function. 1747 1748Yang Tse (11 Mar 2009) 1749- Added missing TELNET timeout support for Windows builds. This issue was 1750 reported by Pierre Brico. 1751 1752Daniel Stenberg (9 Mar 2009) 1753- Frank Hempel found out a bug and provided the fix: 1754 1755 curl_easy_duphandle did not necessarily duplicate the CURLOPT_COOKIEFILE 1756 option. It only enabled the cookie engine in the destination handle if 1757 data->cookies is not NULL (where data is the source handle). In case of a 1758 newly initialized handle which just had the cookie support enabled by a 1759 curl_easy_setopt(handle, CURL_COOKIEFILE, "")-call, handle->cookies was 1760 still NULL because the setopt-call only appends the value to 1761 data->change.cookielist, hence duplicating this handle would not have the 1762 cookie engine switched on. 1763 1764 We also concluded that the slist-functionality would be suitable for being 1765 put in its own module rather than simply hanging out in lib/sendf.c so I 1766 created lib/slist.[ch] for them. 1767 1768- Andreas Farber made the 'buildconf' script check for the presence of m4 1769 scripts to make it detect a bad checkout earlier. People with older 1770 checkouts who don't do cvs update with the -d option won't get the new dirs 1771 and then will get funny outputs that can be a bit hard to understand and 1772 fix. 1773 1774Daniel Stenberg (8 Mar 2009) 1775- Andre Guibert de Bruet found and fixed a code segment in ssluse.c where the 1776 allocation of the memory BIO was not being properly checked. 1777 1778- Andre Guibert de Bruet fixed the gnutls-using code: There are a few places 1779 in the gnutls code where we were checking for negative values for errors, 1780 when the man pages state that GNUTLS_E_SUCCESS is returned on success and 1781 other values indicate error conditions. 1782 1783- Bill Egert pointed out (http://curl.haxx.se/bug/view.cgi?id=2671602) that 1784 curl didn't use sprintf() in a way that is documented to work in POSIX but 1785 since we use our own printf() code (from libcurl) that shouldn't be a 1786 problem. Nonetheless I modified the code to not rely on such particular 1787 features and to not cause further raised eyebrowse with no good reason. 1788 1789Daniel Fandrich (5 Mar 2009) 1790- Expanded the security section of the libcurl-tutorial man page to cover 1791 more issues for authors to consider when writing robust libcurl-using 1792 applications. 1793 1794Yang Tse (5 Mar 2009) 1795- Fixed NTLM authentication memory leak on SSPI enabled Windows builds. This 1796 issue was noticed by Chris Deidun. 1797 1798Daniel Fandrich (4 Mar 2009) 1799- Fixed a problem with m4 quoting in the OpenSSL configure check reported 1800 by Daniel Johnson. 1801 1802Daniel Stenberg (3 Mar 2009) 1803- David James brought a patch that make libcurl close (all) dead connections 1804 whenever you attempt to open a new connection. 1805 1806 1. After cleaning up a dead connection, "continue" instead of 1807 returning FALSE. This ensures that we clean up all dead connections, 1808 rather than just cleaning up the first dead connection. 1809 2. Move up the cleanup for dead connections so that it occurs for 1810 all connections, rather than just the connections which have the same 1811 preferences as our current new connection. 1812 1813Version 7.19.4 (3 March 2009) 1814 1815Daniel Stenberg (3 Mar 2009) 1816- David Kierznowski notified us about a security flaw 1817 (http://curl.haxx.se/docs/adv_20090303.html also known as CVE-2009-0037) in 1818 which previous libcurl versions (by design) can be tricked to access an 1819 arbitrary local/different file instead of a remote one when 1820 CURLOPT_FOLLOWLOCATION is enabled. This flaw is now fixed in this release 1821 together this the addition of two new setopt options for controlling this 1822 new behavior: 1823 1824 o CURLOPT_REDIR_PROTOCOLS controls what protocols libcurl is allowed to 1825 follow to when CURLOPT_FOLLOWLOCATION is enabled. By default, this option 1826 excludes the FILE and SCP protocols and thus you nee to explicitly allow 1827 them in your app if you really want that behavior. 1828 1829 o CURLOPT_PROTOCOLS controls what protocol(s) libcurl is allowed to fetch 1830 using the primary URL option. This is useful if you want to allow a user or 1831 other outsiders control what URL to pass to libcurl and yet not allow all 1832 protocols libcurl may have been built to support. 1833 1834Daniel Stenberg (27 Feb 2009) 1835- Senthil Raja Velu reported a problem when CURLOPT_INTERFACE and 1836 CURLOPT_LOCALPORT were used together (the local port bind failed), and 1837 Markus Koetter provided the fix! 1838 1839Daniel Stenberg (25 Feb 2009) 1840- As Daniel Fandrich figured out, we must do the GnuTLS initing in the 1841 curl_global_init() function to properly maintain the performing functions 1842 thread-safe. We've previously (28 April 2007) moved the init to a later time 1843 just to avoid it to fail very early when libgcrypt dislikes the situation, 1844 but that move was bad and the fix should rather be in libgcrypt or 1845 elsewhere. 1846 1847Daniel Stenberg (24 Feb 2009) 1848- Brian J. Murrell found out that Negotiate proxy authentication didn't work. 1849 It happened because the code used the struct for server-based auth all the 1850 time for both proxy and server auth which of course was wrong. 1851 1852Daniel Stenberg (23 Feb 2009) 1853- After a bug reported by James Cheng I've made curl_easy_getinfo() for 1854 CURLINFO_CONTENT_LENGTH_DOWNLOAD and CURLINFO_CONTENT_LENGTH_UPLOAD return 1855 -1 if the sizes aren't know. Previously these returned 0, make it impossible 1856 to detect the difference between actually zero and unknown. 1857 1858Yang Tse (23 Feb 2009) 1859- Daniel Johnson provided a shell script that will perform all the steps needed 1860 to build a Mac OS X fat ppc/i386 or ppc64/x86_64 libcurl.framework 1861 1862Daniel Stenberg (23 Feb 2009) 1863- I renamed everything in the windows builds files that used the name 'curllib' 1864 to the proper 'libcurl' as clearly this caused confusion. 1865 1866Yang Tse (20 Feb 2009) 1867- Do not halt compilation when using VS2008 to build a Windows 2000 target. 1868 1869Daniel Stenberg (20 Feb 2009) 1870- Linus Nielsen Feltzing reported and helped me repeat and fix a problem with 1871 FTP with the multi interface: when a transfer fails, like when aborted by a 1872 write callback, the control connection was wrongly closed and thus not 1873 re-used properly. 1874 1875 This change is also an attempt to cleanup the code somewhat in this area, as 1876 now the FTP code attempts to keep (better) track on pending responses 1877 necessary to get read in ftp_done(). 1878 1879Daniel Stenberg (19 Feb 2009) 1880- Patrik Thunstrom reported a problem and helped me repeat it. It turned out 1881 libcurl did a superfluous 1000ms wait when doing SFTP downloads! 1882 1883 We read data with libssh2 while doing the "DO" operation for SFTP and then 1884 when we were about to start getting data for the actual file part, the 1885 "TRANSFER" part, we waited for socket action (in 1000ms) before doing a 1886 libssh2-read. But in this case libssh2 had already read and buffered the 1887 data so we ended up always just waiting 1000ms before we get working on the 1888 data! 1889 1890Patrick Monnerat (18 Feb 2009) 1891- FTP downloads (i.e.: RETR) ending with code 550 now return error 1892 CURLE_REMOTE_FILE_NOT_FOUND instead of CURLE_FTP_COULDNT_RETR_FILE. 1893 1894Daniel Stenberg (17 Feb 2009) 1895- Kamil Dudka made NSS-powered builds compile and run again! 1896 1897- A second follow-up change by Andre Guibert de Bruet to fix a related memory 1898 leak like that fixed on the 14th. When zlib returns failure, we need to 1899 cleanup properly before returning error. 1900 1901- CURLOPT_FTP_CREATE_MISSING_DIRS can now be set to 2 in addition to 1 for 1902 plain FTP connections, and it will then allow MKD to fail once and retry the 1903 CWD afterwards. This is especially useful if you're doing many simultanoes 1904 connections against the same server and they all have this option enabled, 1905 as then CWD may first fail but then another connection does MKD before this 1906 connection and thus MKD fails but trying CWD works! The numbers can 1907 (should?) now be set with the convenience enums now called 1908 CURLFTP_CREATE_DIR and CURLFTP_CREATE_DIR_RETRY. 1909 1910 Tests has proven that if you're making an application that uploads a set of 1911 files to an ftp server, you will get a noticable gain in speed if you're 1912 using multiple connections and this option will be then be very useful. 1913 1914Daniel Stenberg (14 Feb 2009) 1915- Andre Guibert de Bruet found and fixed a memory leak in the content encoding 1916 code, which could happen on libz errors. 1917 1918Daniel Fandrich (12 Feb 2009) 1919- Added support for Digest and NTLM authentication using GnuTLS. 1920 1921Daniel Stenberg (11 Feb 2009) 1922- CURLINFO_CONDITION_UNMET was added to allow an application to get to know if 1923 the condition in the previous request was unmet. This is typically a time 1924 condition set with CURLOPT_TIMECONDITION and was previously not possible to 1925 reliably figure out. From bug report #2565128 1926 (http://curl.haxx.se/bug/view.cgi?id=2565128) filed by Jocelyn Jaubert. 1927 1928Daniel Fandrich (4 Feb 2009) 1929- Don't add the standard /usr/lib or /usr/include paths to LDFLAGS and CPPFLAGS 1930 (respectively) when --with-ssl=/usr is used (patch based on FreeBSD). 1931 1932- Added an explicit buffer limit check in msdosify() (patch based on FreeBSD). 1933 This couldn't ever overflow in curl, but might if the code were used 1934 elsewhere or under different conditions. 1935 1936Daniel Stenberg (3 Feb 2009) 1937- Hidemoto Nakada provided a small fix that makes it possible to get the 1938 CURLINFO_CONTENT_LENGTH_DOWNLOAD size from file:// "transfers" with 1939 CURLOPT_NOBODY set true. 1940 1941Daniel Stenberg (2 Feb 2009) 1942- Patrick Scott found a rather large memory leak when using the multi 1943 interface and setting CURLMOPT_MAXCONNECTS to something less than the number 1944 of handles you add to the multi handle. All the connections that didn't fit 1945 in the cache would not be properly disconnected nor freed! 1946 1947- Craig A West brought us: libcurl now defaults to do CONNECT with HTTP 1948 version 1.1 instead of 1.0 like before. This change also introduces the new 1949 proxy type for libcurl called 'CURLPROXY_HTTP_1_0' that then allows apps to 1950 switch (back) to CONNECT 1.0 requests. The curl tool also got a --proxy1.0 1951 option that works exactly like --proxy but sets CURLPROXY_HTTP_1_0. 1952 1953 I updated all test cases cases that use CONNECT and I tried to do some using 1954 --proxy1.0 and some updated to do CONNECT 1.1 to get both versions run. 1955 1956Daniel Stenberg (31 Jan 2009) 1957- When building with c-ares 1.6.1 (not yet released) or later and IPv6 support 1958 enabled, we can now take advantage of its brand new AF_UNSPEC support in 1959 ares_gethostbyname(). This makes test case 241 finally run fine for me with 1960 this setup since it now parses the "::1 ip6-localhost" line fine in my 1961 /etc/hosts file! 1962 1963Daniel Stenberg (30 Jan 2009) 1964- Scott Cantor filed bug report #2550061 1965 (http://curl.haxx.se/bug/view.cgi?id=2550061) mentioning that I failed to 1966 properly make sure that the VC9 makefiles got included in the latest 1967 release. I've now fixed the release script and verified it so next release 1968 will hopefully include them properly! 1969 1970Daniel Fandrich (30 Jan 2009) 1971- Fixed --disable-proxy for FTP and SOCKS. Thanks to Daniel Egger for 1972 reporting. 1973 1974Yang Tse (29 Jan 2009) 1975- Introduced curl_sspi.c and curl_sspi.h for the implementation of functions 1976 Curl_sspi_global_init() and Curl_sspi_global_cleanup() which previously were 1977 named Curl_ntlm_global_init() and Curl_ntlm_global_cleanup() in http_ntlm.c 1978 Also adjusted socks_sspi.c to remove the link-time dependency on the Windows 1979 SSPI library using it now in the same way as it was done in http_ntlm.c. 1980 1981Daniel Stenberg (28 Jan 2009) 1982- Markus Moeller introduced two new options to libcurl: 1983 CURLOPT_SOCKS5_GSSAPI_SERVICE and CURLOPT_SOCKS5_GSSAPI_NEC to allow libcurl 1984 to do GSS-style authentication with SOCKS5 proxies. The curl tool got the 1985 options called --socks5-gssapi-service and --socks5-gssapi-nec to enable 1986 these. 1987 1988Daniel Stenberg (26 Jan 2009) 1989- Chad Monroe provided the new CURLOPT_TFTP_BLKSIZE option that allows an app 1990 to set desired block size to use for TFTP transfers instead of the default 1991 512 bytes. 1992 1993- The "-no_ticket" option was introduced in Openssl0.9.8j. It's a flag to 1994 disable "rfc4507bis session ticket support". rfc4507bis was later turned 1995 into the proper RFC5077 it seems: http://tools.ietf.org/html/rfc5077 1996 1997 The enabled extension concerns the session management. I wonder how often 1998 libcurl stops a connection and then resumes a TLS session. also, sending the 1999 session data is some overhead. .I suggest that you just use your proposed 2000 patch (which explicitly disables TICKET). 2001 2002 If someone writes an application with libcurl and openssl who wants to 2003 enable the feature, one can do this in the SSL callback. 2004 2005 Sharad Gupta brought this to my attention. Peter Sylvester helped me decide 2006 on the proper action. 2007 2008- Alexey Borzov filed bug report #2535504 2009 (http://curl.haxx.se/bug/view.cgi?id=2535504) pointing out that realms with 2010 quoted quotation marks in HTTP Digest headers didn't work. I've now added 2011 test case 1095 that verifies my fix. 2012 2013- Craig A West brought CURLOPT_NOPROXY and the corresponding --noproxy option. 2014 They basically offer the same thing the NO_PROXY environment variable only 2015 offered previously: list a set of host names that shall not use the proxy 2016 even if one is specified. 2017 2018Daniel Fandrich (20 Jan 2009) 2019- Call setlocale() for libtest tests to test the effects of locale-induced 2020 libc changes on libcurl. 2021 2022- Fixed a couple more locale-dependent toupper conversions, mainly for 2023 clarity. This does fix one problem that causes ;type=i FTP URLs 2024 to fail in the Turkish locale when CURLOPT_PROXY_TRANSFER_MODE is 2025 used (test case 561) 2026 2027- Added tests 561 and 1091 through 1094 to test various combinations 2028 of ;type= and ;mode= URLs that could potentially fail in the Turkish 2029 locale. 2030 2031Daniel Stenberg (20 Jan 2009) 2032- Lisa Xu pointed out that the ssh.obj file was missing from the 2033 lib/Makefile.vc6 file (and thus from the vc8 and vc9 ones too). 2034 2035Version 7.19.3 (19 January 2009) 2036 2037Daniel Stenberg (16 Jan 2009) 2038- Andrew de los Reyes fixed curlbuild.h for "generic" gcc builds on PPC, both 2039 32 bit and 64 bit. 2040 2041Daniel Stenberg (15 Jan 2009) 2042- Tim Ansell fixed a compiler warning in lib/cookie.c 2043 2044Daniel Stenberg (14 Jan 2009) 2045- Grant Erickson fixed timeouts for TFTP such that specifying a 2046 connect-timeout, a max-time or both options work correctly and as expected 2047 by passing the correct boolean value to Curl_timeleft via the 2048 'duringconnect' parameter. 2049 2050 With this small change, curl TFTP now behaves as expected (and likely as 2051 originally-designed): 2052 2053 1) For non-existent or unreachable dotted IP addresses: 2054 2055 a) With no options, follows the default curl 300s timeout... 2056 b) With --connect-timeout only, follows that value... 2057 c) With --max-time only, follows that value... 2058 d) With both --connect-timeout and --max-time, follows the smaller value... 2059 2060 and times out with a "curl: (7) Couldn't connect to server" error. 2061 2062 2) For transfers to/from a valid host: 2063 2064 a) With no options, follows default curl 300s timeout for the 2065 first XRQ/DATA/ACK transaction and the default TFTP 3600s 2066 timeout for the remainder of the transfer... 2067 2068 b) With --connect-time only, follows that value for the 2069 first XRQ/DATA/ACK transaction and the default TFTP 3600s 2070 timeout for the remainder of the transfer... 2071 2072 c) With --max-time only, follows that value for the first 2073 XRQ/DATA/ACK transaction and for the remainder of the 2074 transfer... 2075 2076 d) With both --connect-timeout and --max-time, follows the former 2077 for the first XRQ/DATA/ACK transaction and the latter for the 2078 remainder of the transfer... 2079 2080 and times out with a "curl: (28) Timeout was reached" error as 2081 appropriate. 2082 2083Daniel Stenberg (13 Jan 2009) 2084- Michael Wallner fixed a NULL pointer deref when calling 2085 curl_easy_setup(curl, CURLOPT_COOKIELIST, "SESS") on a CURL handle with no 2086 cookies data. 2087 2088- Stefan Teleman brought a patch to fix the default curlbuild.h file for the 2089 SunPro compilers. 2090 2091Daniel Stenberg (12 Jan 2009) 2092- Based on bug report #2498665 (http://curl.haxx.se/bug/view.cgi?id=2498665) 2093 by Daniel Black, I've now added magic to the configure script that makes it 2094 use pkg-config to detect gnutls details as well if the existing method 2095 (using libgnutls-config) fails. While doing this, I cleaned up and unified 2096 the pkg-config usage when detecting openssl and nss as well. 2097 2098Daniel Stenberg (11 Jan 2009) 2099- Karl Moerder brought the patch that creates vc9 Makefiles, and I made 2100 'maketgz' now use the actual makefile targets to do the VC8 and VC9 2101 makefiles. 2102 2103Daniel Stenberg (10 Jan 2009) 2104- Emil Romanus fixed: 2105 2106 When using the multi interface over HTTP and the server returns a Location 2107 header, the running easy handle will get stuck in the CURLM_STATE_PERFORM 2108 state, leaving the external event loop stuck waiting for data from the 2109 ingoing socket (when using the curl_multi_socket_action stuff). While this 2110 bug was pretty hard to find, it seems to require only a one-line fix. The 2111 break statement on line 1374 in multi.c caused the function to skip the call 2112 to multistate(). 2113 2114 How to reproduce this bug? Well, that's another question. evhiperfifo.c in 2115 the examples directory chokes on this bug only _sometimes_, probably 2116 depending on how fast the URLs are added. One way of testing the bug out is 2117 writing to hiper.fifo from more than one source at the same time. 2118 2119Daniel Fandrich (7 Jan 2009) 2120- Unified much of the SessionHandle initialization done in Curl_open() and 2121 curl_easy_reset() by creating Curl_init_userdefined(). This had the side 2122 effect of fixing curl_easy_reset() so it now also resets 2123 CURLOPT_FTP_FILEMETHOD and CURLOPT_SSL_SESSIONID_CACHE 2124 2125Daniel Stenberg (7 Jan 2009) 2126- Rob Crittenden did once again provide an NSS update: 2127 2128 I have to jump through a few hoops now with the NSS library initialization 2129 since another part of an application may have already initialized NSS by the 2130 time Curl gets invoked. This patch is more careful to only shutdown the NSS 2131 library if Curl did the initialization. 2132 2133 It also adds in a bit of code to set the default ciphers if the app that 2134 call NSS_Init* did not call NSS_SetDomesticPolicy() or set specific 2135 ciphers. One might argue that this lets other application developers get 2136 lazy and/or they aren't using the NSS API correctly, and you'd be right. 2137 But still, this will avoid terribly difficult-to-trace crashes and is 2138 generally helpful. 2139 2140Daniel Stenberg (1 Jan 2009) 2141- 'reconf' is removed since we rather have users use 'buildconf' 2142 2143Daniel Stenberg (31 Dec 2008) 2144- Bas Mevissen reported http://curl.haxx.se/bug/view.cgi?id=2479030 pointing 2145 out that 'reconf' didn't properly point out the m4 subdirectory when running 2146 aclocal. 2147 2148Daniel Stenberg (29 Dec 2008) 2149 - Phil Lisiecki filed bug report #2413067 2150 (http://curl.haxx.se/bug/view.cgi?id=2413067) that identified a problem that 2151 would cause libcurl to mark a DNS cache entry "in use" eternally if the 2152 subsequence TCP connect failed. It would thus never get pruned and refreshed 2153 as it should've been. 2154 2155 Phil provided his own patch to this problem that while it seemed to work 2156 wasn't complete and thus I wrote my own fix to the problem. 2157 2158Daniel Stenberg (28 Dec 2008) 2159- Peter Korsgaard fixed building libcurl with "configure --with-ssl 2160 --disable-verbose". 2161 2162- Anthony Bryan fixed more language and spelling flaws in man pages. 2163 2164Daniel Stenberg (22 Dec 2008) 2165- Given a recent enough libssh2, libcurl can now seek/resume with SFTP even 2166 on file indexes beyond 2 or 4GB. 2167 2168- Anthony Bryan provided a set of patches that cleaned up manual language, 2169 corrected spellings and more. 2170 2171Daniel Stenberg (20 Dec 2008) 2172- Igor Novoseltsev fixed a bad situation for the multi_socket() API when doing 2173 pipelining, as libcurl could then easily get confused and A) work on the 2174 handle that was not "first in queue" on a pipeline, or even B) tell the app 2175 to REMOVE a socket while it was in use by a second handle in a pipeline. Both 2176 errors caused hanging or stalling applications. 2177 2178Daniel Stenberg (19 Dec 2008) 2179- curl_multi_timeout() could return a timeout value of 0 even though nothing 2180 was actually ready to get done, as the internal time resolution is higher 2181 than the returned millisecond timer. Therefore it could cause applications 2182 running on fast processors to do short bursts of busy-loops. 2183 curl_multi_timeout() will now only return 0 if the timeout is actually 2184 alreay triggered. 2185 2186- Using the libssh2 0.19 function libssh2_session_block_directions(), libcurl 2187 now has an improved ability to do right when the multi interface (both 2188 "regular" and multi_socket) is used for SCP and SFTP transfers. This should 2189 result in (much) less busy-loop situations and thus less CPU usage with no 2190 speed loss. 2191 2192Daniel Stenberg (17 Dec 2008) 2193- SCP and SFTP with the multi interface had the same flaw: the 'DONE' 2194 operation didn't complete properly if the EAGAIN equivalent was returned but 2195 libcurl would simply continue with a half-completed close operation 2196 performed. This ruined persistent connection re-use and cause some 2197 SSH-protocol errors in general. The correction is unfortunately adding a 2198 blocking function - doing it entirely non-blocking should be considered for 2199 a better fix. 2200 2201Gisle Vanem (16 Dec 2008) 2202- Added the possibility to use the Watt-32 tcp/ip stack under Windows. 2203 The change simply involved adding a USE_WATT32 section in the 2204 config-win32.h files (under ./lib and ./src). This section disables 2205 the use of any Winsock headers. 2206 2207Daniel Stenberg (16 Dec 2008) 2208- libssh2_sftp_last_error() was wrongly used at some places in libcurl which 2209 made libcurl sometimes not properly abort problematic SFTP transfers. 2210 2211Daniel Stenberg (12 Dec 2008) 2212- More work with Igor Novoseltsev to first fix the remaining stuff for 2213 removing easy handles from multi handles when the easy handle is/was within 2214 a HTTP pipeline. His bug report #2351653 2215 (http://curl.haxx.se/bug/view.cgi?id=2351653) was also related and was 2216 eventually fixed by a patch by Igor himself. 2217 2218Yang Tse (12 Dec 2008) 2219- Patrick Monnerat fixed a build regression, introduced in 7.19.2, affecting 2220 OS/400 compilations with IPv6 enabled. 2221 2222Daniel Stenberg (12 Dec 2008) 2223- Mark Karpeles filed bug report #2416182 titled "crash in ConnectionExists 2224 when using duphandle+curl_mutli" 2225 (http://curl.haxx.se/bug/view.cgi?id=2416182) which showed that 2226 curl_easy_duphandle() wrongly also copied the pointer to the connection 2227 cache, which was plain wrong and caused a segfault if the handle would be 2228 used in a different multi handle than the handle it was duplicated from. 2229 2230Daniel Stenberg (11 Dec 2008) 2231- Keshav Krity found out that libcurl failed to deal with dotted IPv6 2232 addresses if they were very long (>39 letters) due to a too strict address 2233 validity parser. It now accepts addresses up to 45 bytes long. 2234 2235Daniel Stenberg (11 Dec 2008) 2236- Internet Explorer had a broken HTTP digest authentication before v7 and 2237 there are servers "out there" that relies on the client doing this broken 2238 Digest authentication. Apache even comes with an option to work with such 2239 broken clients. 2240 2241 The difference is only for URLs that contain a query-part (a '?'-letter and 2242 text to the right of it). 2243 2244 libcurl now supports this quirk, and you enable it by setting the 2245 CURLAUTH_DIGEST_IE bit in the bitmask you pass to the CURLOPT_HTTPAUTH or 2246 CURLOPT_PROXYAUTH options. They are thus individually controlled to server 2247 and proxy. 2248 2249 (note that there's no way to activate this with the curl tool yet) 2250 2251Daniel Fandrich (9 Dec 2008) 2252- Added test cases 1089 and 1090 to test --write-out after a redirect to 2253 test a report that the size didn't work, but these test cases pass. 2254 2255- Documented CURLOPT_CONNECT_ONLY as being useful only on HTTP URLs. 2256 2257Daniel Stenberg (9 Dec 2008) 2258- Ken Hirsch simplified how libcurl does FTPS: now it doesn't assume any 2259 particular state for the control connection like it did before for implicit 2260 FTPS (libcurl assumed such control connections to be encrypted while some 2261 FTPS servers such as FileZilla assumes such connections to be clear 2262 mode). Use the CURLOPT_USE_SSL option to set your desired level. 2263 2264Daniel Stenberg (8 Dec 2008) 2265- Fred Machado posted about a weird FTP problem on the curl-users list and when 2266 researching it, it turned out he got a 550 response back from a SIZE command 2267 and then I fell over the text in RFC3659 that says: 2268 2269 The presence of the 550 error response to a SIZE command MUST NOT be taken 2270 by the client as an indication that the file cannot be transferred in the 2271 current MODE and TYPE. 2272 2273 In other words: the change I did on September 30th 2008 and that has been 2274 included in the last two releases were a regression and a bad idea. We MUST 2275 NOT take a 550 response from SIZE as a hint that the file doesn't exist. 2276 2277- Christian Krause filed bug #2221237 2278 (http://curl.haxx.se/bug/view.cgi?id=2221237) that identified an infinite 2279 loop during GSS authentication given some specific conditions. With his 2280 patience and great feedback I managed to narrow down the problem and 2281 eventually fix it although I can't test any of this myself! 2282 2283Daniel Fandrich (3 Dec 2008) 2284- Fixed the getifaddrs version of Curl_if2ip to work on systems without IPv6 2285 support (e.g. Minix) 2286 2287Daniel Stenberg (3 Dec 2008) 2288- Igor Novoseltsev filed bug #2351645 2289 (http://curl.haxx.se/bug/view.cgi?id=2351645) that identified a problem with 2290 the multi interface that occured if you removed an easy handle while in 2291 progress and the handle was used in a HTTP pipeline. 2292 2293- Pawel Kierski pointed out a mistake in the cookie code that could lead to a 2294 bad fclose() after a fatal error had occured. 2295 (http://curl.haxx.se/bug/view.cgi?id=2382219) 2296 2297Daniel Fandrich (25 Nov 2008) 2298- If a HTTP request is Basic and num is already >=1000, the HTTP test 2299 server adds 1 to num to get the data section to return. This allows 2300 testing authentication negotiations using the Basic authentication 2301 method. 2302 2303- Added tests 1087 and 1088 to test Basic authentication on a redirect 2304 with and without --location-trusted 2305 2306Daniel Stenberg (24 Nov 2008) 2307- Based on a patch by Vlad Grachov, libcurl now uses a new libssh2 0.19 2308 function when built to support SCP and SFTP that helps the library to know 2309 in which direction a particular libssh2 operation would return EAGAIN so 2310 that libcurl knows what socket conditions to wait for before trying the 2311 function call again. Previously (and still when using libssh2 0.18 or 2312 earlier), libcurl will busy-loop in this situation when the easy interface 2313 is used! 2314 2315Daniel Fandrich (20 Nov 2008) 2316- Automatically detect OpenBSD's CA cert bundle. 2317 2318Daniel Stenberg (19 Nov 2008) 2319- I removed the default use of "Pragma: no-cache" from libcurl when a proxy is 2320 used. It has been used since forever but it was never a good idea to use 2321 unless explicitly asked for. 2322 2323- Josef Wolf's extension that allows a $TESTDIR/gdbinit$testnum file that when 2324 you use runtests.pl -g, will be sourced by gdb to allow additional fancy or 2325 whatever you see fit 2326 2327- Christian Krause reported and fixed a memory leak that would occur with HTTP 2328 GSS/kerberos authentication (http://curl.haxx.se/bug/view.cgi?id=2284386) 2329 2330- Andreas Wurf and Markus Koetter helped me analyze a problem that Andreas got 2331 when uploading files to a single FTP server using multiple easy handle 2332 handles with the multi interface. Occasionally a handle would stall in 2333 mysterious ways. 2334 2335 The problem turned out to be a side-effect of the ConnectionExists() 2336 function's eagerness to re-use a handle for HTTP pipelining so it would 2337 select it even if already being in use, due to an inadequate check for its 2338 chances of being used for pipelnining. 2339 2340Daniel Fandrich (17 Nov 2008) 2341- Added more compiler warning options for gcc 4.3 2342 2343Yang Tse (17 Nov 2008) 2344- Fix a remaining problem in the inet_pton() runtime configure check. And 2345 fix internal Curl_inet_pton() failures to reject certain malformed literals. 2346 2347- Make configure script check if ioctl with the SIOCGIFADDR command can be 2348 used, and define HAVE_IOCTL_SIOCGIFADDR if appropriate. 2349 2350Daniel Stenberg (16 Nov 2008) 2351- Christian Krause fixed a build failure when building with gss support 2352 enabled and FTP disabled. 2353 2354- Added check for NULL returns from strdup() in src/main.c and lib/formdata.c 2355 - reported by Jim Meyering also prevent buffer overflow on MSDOS when you do 2356 for example -O on a url with a file name part longer than PATH_MAX letters 2357 2358- lib/nss.c fixes based on the report by Jim Meyering: I went over and added 2359 checks for return codes for all calls to malloc and strdup that were 2360 missing. I also changed a few malloc(13) to use arrays on the stack and a 2361 few malloc(PATH_MAX) to instead use aprintf() to lower memory use. 2362 2363- I fixed a memory leak in Curl_nss_connect() when CURLOPT_ISSUERCERT is 2364 in use. 2365 2366Daniel Fandrich (14 Nov 2008) 2367- Added .xml as one of the few common file extensions known by the multipart 2368 form generator. 2369 2370- Added some #ifdefs around header files and change the EAGAIN test to 2371 fix compilation on Cell (reported by Jeff Curley). 2372 2373Yang Tse (14 Nov 2008) 2374- Fixed several configure script issues affecting checks for inet_ntoa_r(), 2375 inet_ntop(), inet_pton(), getifaddrs(), fcntl() and getaddrinfo(). 2376 2377Yang Tse (13 Nov 2008) 2378- Refactored configure script detection of functions used to set sockets into 2379 non-blocking mode, and decouple function detection from function capability. 2380 2381