5 \___|\___/|_| \_\_____|
9 These are problems and bugs known to exist at the time of this release. Feel
10 free to join in and help us correct one or more of these! Also be sure to
11 check the changelog of the current development status, as one or more of these
12 problems may have been fixed or changed somewhat since this was written!
15 1.1 CURLFORM_CONTENTLEN in an array
16 1.2 Disabling HTTP Pipelining
17 1.3 STARTTRANSFER time is wrong for HTTP POSTs
18 1.4 multipart formposts file name encoding
19 1.5 Expect-100 meets 417
20 1.6 Unnecessary close when 401 received waiting for 100
21 1.8 DNS timing is wrong for HTTP redirects
22 1.9 HTTP/2 frames while in the connection pool kill reuse
23 1.10 Strips trailing dot from host name
24 1.11 CURLOPT_SEEKFUNCTION not called with CURLFORM_STREAM
27 2.1 CURLINFO_SSL_VERIFYRESULT has limited support
29 2.3 GnuTLS backend skips really long certificate fields
32 3.1 IMAP SEARCH ALL truncated response
33 3.2 No disconnect command
34 3.3 SMTP to multiple recipients
35 3.4 POP3 expects "CRLF.CRLF" eob for some single-line responses
38 4.1 -J with %-encoded file nameas
39 4.2 -J with -C - fails
40 4.3 --retry and transfer timeouts
42 5. Build and portability issues
43 5.1 Windows Borland compiler
44 5.2 curl-config --libs contains private details
45 5.3 libidn and old iconv
46 5.4 AIX shared build with c-ares fails
47 5.5 can't handle Unicode arguments in Windows
48 5.6 cmake support gaps
49 5.7 Visual Studio project gaps
50 5.8 configure finding libs in wrong directory
51 5.9 Utilize Requires.private directives in libcurl.pc
52 5.10 Fix the gcc typechecks
55 6.1 NTLM authentication and unicode
56 6.2 MIT Kerberos for Windows build
57 6.3 NTLM in system context uses wrong name
58 6.4 Negotiate and Kerberos V5 need a fake user name
61 7.1 FTP without or slow 220 response
62 7.2 FTP with CONNECT and slow server
63 7.3 FTP with NOBODY and FAILONERROR
66 7.6 FTP with NULs in URL parts
67 7.7 FTP and empty path parts in the URL
68 7.8 Premature transfer end but healthy control channel
71 8.1 TELNET and time limtiations don't work
72 8.2 Microsoft telnet server
75 9.1 SFTP doesn't do CURLOPT_POSTQUOTE correct
78 10.1 SOCKS proxy connections are done blocking
79 10.2 SOCKS don't support timeouts
81 10.4 active FTP over a SOCKS
84 11.1 Curl leaks .onion hostnames in DNS
85 11.2 error buffer not set if connection to multiple addresses fails
86 11.3 c-ares deviates from stock resolver on http://1346569778
89 12.1 OpenLDAP hangs after returning results
92 13.1 --interface for ipv6 binds to unusable IP address
95 ==============================================================================
99 1.1 CURLFORM_CONTENTLEN in an array
101 It is not possible to pass a 64-bit value using CURLFORM_CONTENTLEN with
102 CURLFORM_ARRAY, when compiled on 32-bit platforms that support 64-bit
103 integers. This is because the underlying structure 'curl_forms' uses a dual
104 purpose char* for storing these values in via casting. For more information
105 see the now closed related issue:
106 https://github.com/curl/curl/issues/608
108 1.2 Disabling HTTP Pipelining
110 Disabling HTTP Pipelining when there are ongoing transfers can lead to
111 heap corruption and crash. https://curl.haxx.se/bug/view.cgi?id=1411
113 1.3 STARTTRANSFER time is wrong for HTTP POSTs
115 Wrong STARTTRANSFER timer accounting for POST requests Timer works fine with
116 GET requests, but while using POST the time for CURLINFO_STARTTRANSFER_TIME
117 is wrong. While using POST CURLINFO_STARTTRANSFER_TIME minus
118 CURLINFO_PRETRANSFER_TIME is near to zero every time.
120 https://github.com/curl/curl/issues/218
121 https://curl.haxx.se/bug/view.cgi?id=1213
123 1.4 multipart formposts file name encoding
125 When creating multipart formposts. The file name part can be encoded with
126 something beyond ascii but currently libcurl will only pass in the verbatim
127 string the app provides. There are several browsers that already do this
128 encoding. The key seems to be the updated draft to RFC2231:
129 https://tools.ietf.org/html/draft-reschke-rfc2231-in-http-02
131 1.5 Expect-100 meets 417
133 If an upload using Expect: 100-continue receives an HTTP 417 response, it
134 ought to be automatically resent without the Expect:. A workaround is for
135 the client application to redo the transfer after disabling Expect:.
136 https://curl.haxx.se/mail/archive-2008-02/0043.html
138 1.6 Unnecessary close when 401 received waiting for 100
140 libcurl closes the connection if an HTTP 401 reply is received while it is
141 waiting for the the 100-continue response.
142 https://curl.haxx.se/mail/lib-2008-08/0462.html
144 1.8 DNS timing is wrong for HTTP redirects
146 When extracting timing information after HTTP redirects, only the last
147 transfer's results are returned and not the totals:
148 https://github.com/curl/curl/issues/522
150 1.9 HTTP/2 frames while in the connection pool kill reuse
152 If the server sends HTTP/2 frames (like for example an HTTP/2 PING frame) to
153 curl while the connection is held in curl's connection pool, the socket will
154 be found readable when considered for reuse and that makes curl think it is
155 dead and then it will be closed and a new connection gets created instead.
157 This is *best* fixed by adding monitoring to connections while they are kept
158 in the pool so that pings can be responded to appropriately.
160 1.10 Strips trailing dot from host name
162 When given a URL wit a trailing dot for the host name part:
163 "https://example.com./", libcurl will strip off the dot and use the name
164 without a dot internally and send it dot-less in HTTP Host: headers and in
167 The HTTP part violates RFC 7230 section 5.4 but the SNI part is accordance
168 with RFC 6066 section 3.
170 URLs using these trailing dots are very rare in the wild and we have not seen
171 or gotten any real-world problems with such URLs reported. The popular
172 browsers seem to have stayed with not stripping the dot for both uses (thus
173 they violate RFC 6066 instead of RFC 7230).
175 Daniel took the discussion to the HTTPbis mailing list in March 2016:
176 https://lists.w3.org/Archives/Public/ietf-http-wg/2016JanMar/0430.html but
177 there was not major rush or interest to fix this. The impression I get is
178 that most HTTP people rather not rock the boat now and instead prioritize web
179 compatibility rather than to strictly adhere to these RFCs.
181 Our current approach allows a knowing client to send a custom HTTP header
184 It can also be noted that while adding a trailing dot to the host name in
185 most (all?) cases will make the name resolve to the same set of IP addresses,
186 many HTTP servers will not happily accept the trailing dot there unless that
187 has been specificly configured to be a fine virtual host.
189 If URLs with trailing dots for host names become more popular or even just
190 used more than for just plain fun experiments, I'm sure we will have reason
191 to go back and reconsider.
193 See https://github.com/curl/curl/issues/716 for the discussion.
195 1.11 CURLOPT_SEEKFUNCTION not called with CURLFORM_STREAM
197 I'm using libcurl to POST form data using a FILE* with the CURLFORM_STREAM
198 option of curl_formadd(). I've noticed that if the connection drops at just
199 the right time, the POST is reattempted without the data from the file. It
200 seems like the file stream position isn't getting reset to the beginning of
201 the file. I found the CURLOPT_SEEKFUNCTION option and set that with a
202 function that performs an fseek() on the FILE*. However, setting that didn't
203 seem to fix the issue or even get called. See
204 https://github.com/curl/curl/issues/768
209 2.1 CURLINFO_SSL_VERIFYRESULT has limited support
211 CURLINFO_SSL_VERIFYRESULT is only implemented for the OpenSSL and NSS
212 backends, so relying on this information in a generic app is flaky.
216 Curl doesn't recognize certificates in DER format in keychain, but it works
217 with PEM. https://curl.haxx.se/bug/view.cgi?id=1065
219 2.3 GnuTLS backend skips really long certificate fields
221 libcurl calls gnutls_x509_crt_get_dn() with a fixed buffer size and if the
222 field is too long in the cert, it'll just return an error and the field will
228 3.1 IMAP SEARCH ALL truncated response
230 IMAP "SEARCH ALL" truncates output on large boxes. "A quick search of the
231 code reveals that pingpong.c contains some truncation code, at line 408, when
232 it deems the server response to be too large truncating it to 40 characters"
233 https://curl.haxx.se/bug/view.cgi?id=1366
235 3.2 No disconnect command
237 The disconnect commands (LOGOUT and QUIT) may not be sent by IMAP, POP3 and
238 SMTP if a failure occurs during the authentication phase of a connection.
240 3.3 SMTP to multiple recipients
242 When sending data to multiple recipients, curl will abort and return failure
243 if one of the recipients indicate failure (on the "RCPT TO"
244 command). Ordinary mail programs would proceed and still send to the ones
245 that can receive data. This is subject for change in the future.
246 https://curl.haxx.se/bug/view.cgi?id=1116
248 3.4 POP3 expects "CRLF.CRLF" eob for some single-line responses
250 You have to tell libcurl not to expect a body, when dealing with one line
251 response commands. Please see the POP3 examples and test cases which show
252 this for the NOOP and DELE commands. https://curl.haxx.se/bug/?i=740
257 4.1 -J with %-encoded file nameas
259 -J/--remote-header-name doesn't decode %-encoded file names. RFC6266 details
260 how it should be done. The can of worm is basically that we have no charset
261 handling in curl and ascii >=128 is a challenge for us. Not to mention that
262 decoding also means that we need to check for nastiness that is attempted,
263 like "../" sequences and the like. Probably everything to the left of any
264 embedded slashes should be cut off.
265 https://curl.haxx.se/bug/view.cgi?id=1294
267 4.2 -J with -C - fails
269 When using -J (with -O), automatically resumed downloading together with "-C
270 -" fails. Without -J the same command line works! This happens because the
271 resume logic is worked out before the target file name (and thus its
272 pre-transfer size) has been figured out!
273 https://curl.haxx.se/bug/view.cgi?id=1169
275 4.3 --retry and transfer timeouts
277 If using --retry and the transfer timeouts (possibly due to using -m or
278 -y/-Y) the next attempt doesn't resume the transfer properly from what was
279 downloaded in the previous attempt but will truncate and restart at the
280 original position where it was at before the previous failed attempt. See
281 https://curl.haxx.se/mail/lib-2008-01/0080.html and Mandriva bug report
282 https://qa.mandriva.com/show_bug.cgi?id=22565
285 5. Build and portability issues
287 5.1 Windows Borland compiler
289 When building with the Windows Borland compiler, it fails because the "tlib"
290 tool doesn't support hyphens (minus signs) in file names and we have such in
291 the build. https://curl.haxx.se/bug/view.cgi?id=1222
293 5.2 curl-config --libs contains private details
295 "curl-config --libs" will include details set in LDFLAGS when configure is
296 run that might be needed only for building libcurl. Further, curl-config
297 --cflags suffers from the same effects with CFLAGS/CPPFLAGS.
299 5.3 libidn and old iconv
301 Test case 165 might fail on a system which has libidn present, but with an
302 old iconv version (2.1.3 is a known bad version), since it doesn't recognize
303 the charset when named ISO8859-1. Changing the name to ISO-8859-1 makes the
304 test pass, but instead makes it fail on Solaris hosts that use its native
307 5.4 AIX shared build with c-ares fails
309 curl version 7.12.2 fails on AIX if compiled with --enable-ares. The
310 workaround is to combine --enable-ares with --disable-shared
312 5.5 can't handle Unicode arguments in Windows
314 If a URL or filename can't be encoded using the user's current codepage then
315 it can only be encoded properly in the Unicode character set. Windows uses
316 UTF-16 encoding for Unicode and stores it in wide characters, however curl
317 and libcurl are not equipped for that at the moment. And, except for Cygwin,
318 Windows can't use UTF-8 as a locale.
320 https://curl.haxx.se/bug/?i=345
321 https://curl.haxx.se/bug/?i=731
323 5.6 cmake support gaps
325 The cmake build setup lacks several features that the autoconf build
326 offers. This includes:
328 - symbol hiding when the shared library is built
329 - use of correct soname for the shared library build
330 - support for several TLS backends are missing
331 - the unit tests cause link failures in regular non-static builds
334 5.7 Visual Studio project gaps
336 The Visual Studio projects lack some features that the autoconf and nmake
337 builds offer, such as the following:
339 - support for zlib and nghttp2
340 - use of static runtime libraries
341 - add the test suite components
343 In addition to this the following could be implemented:
345 - support for other development IDEs
346 - add PATH environment variables for third-party DLLs
348 5.8 configure finding libs in wrong directory
350 When the configure script checks for third-party libraries, it adds those
351 directories to the LDFLAGS variable and then tries linking to see if it
352 works. When successful, the found directory is kept in the LDFLAGS variable
353 when the script continues to execute and do more tests and possibly check for
356 This can make subsequent checks for libraries wrongly detect another
357 installation in a directory that was previously added to LDFLAGS by another
360 A possibly better way to do these checks would be to keep the pristine LDFLAGS
361 even after successful checks and instead add those verified paths to a
362 separate variable that only after all library checks have been performed gets
365 5.9 Utilize Requires.private directives in libcurl.pc
367 https://github.com/curl/curl/issues/864
369 5.10 Fix the gcc typechecks
371 Issue #846 identifies a problem with the gcc-typechecks and how the types are
372 documented and checked for CURLINFO_CERTINFO but our attempts to fix the
373 issue were futile and needs more attention.
375 https://github.com/curl/curl/issues/846
379 6.1 NTLM authentication and unicode
381 NTLM authentication involving unicode user name or password only works
382 properly if built with UNICODE defined together with the WinSSL/schannel
383 backend. The original problem was mentioned in:
384 https://curl.haxx.se/mail/lib-2009-10/0024.html
385 https://curl.haxx.se/bug/view.cgi?id=896
387 The WinSSL/schannel version verified to work as mentioned in
388 https://curl.haxx.se/mail/lib-2012-07/0073.html
390 6.2 MIT Kerberos for Windows build
392 libcurl fails to build with MIT Kerberos for Windows (KfW) due to KfW's
393 library header files exporting symbols/macros that should be kept private to
394 the KfW library. See ticket #5601 at http://krbdev.mit.edu/rt/
396 6.3 NTLM in system context uses wrong name
398 NTLM authentication using SSPI (on Windows) when (lib)curl is running in
399 "system context" will make it use wrong(?) user name - at least when compared
400 to what winhttp does. See https://curl.haxx.se/bug/view.cgi?id=535
402 6.4 Negotiate and Kerberos V5 need a fake user name
404 In order to get Negotiate (SPNEGO) authentication to work in HTTP or Kerberos
405 V5 in the e-mail protocols, you need to provide a (fake) user name (this
406 concerns both curl and the lib) because the code wrongly only considers
407 authentication if there's a user name provided by setting
408 conn->bits.user_passwd in url.c https://curl.haxx.se/bug/view.cgi?id=440 How?
409 https://curl.haxx.se/mail/lib-2004-08/0182.html A possible solution is to
410 either modify this variable to be set or introduce a variable such as
411 new conn->bits.want_authentication which is set when any of the authentication
417 7.1 FTP without or slow 220 response
419 If a connection is made to a FTP server but the server then just never sends
420 the 220 response or otherwise is dead slow, libcurl will not acknowledge the
421 connection timeout during that phase but only the "real" timeout - which may
422 surprise users as it is probably considered to be the connect phase to most
423 people. Brought up (and is being misunderstood) in:
424 https://curl.haxx.se/bug/view.cgi?id=856
426 7.2 FTP with CONNECT and slow server
428 When doing FTP over a socks proxy or CONNECT through HTTP proxy and the multi
429 interface is used, libcurl will fail if the (passive) TCP connection for the
430 data transfer isn't more or less instant as the code does not properly wait
431 for the connect to be confirmed. See test case 564 for a first shot at a test
434 7.3 FTP with NOBODY and FAILONERROR
436 It seems sensible to be able to use CURLOPT_NOBODY and CURLOPT_FAILONERROR
437 with FTP to detect if a file exists or not, but it is not working:
438 https://curl.haxx.se/mail/lib-2008-07/0295.html
442 When doing an operation over FTP that requires the ACCT command (but not when
443 logging in), the operation will fail since libcurl doesn't detect this and
444 thus fails to issue the correct command:
445 https://curl.haxx.se/bug/view.cgi?id=635
449 FTP ASCII transfers do not follow RFC959. They don't convert the data
450 accordingly (not for sending nor for receiving). RFC 959 section 3.1.1.1
451 clearly describes how this should be done:
453 The sender converts the data from an internal character representation to
454 the standard 8-bit NVT-ASCII representation (see the Telnet
455 specification). The receiver will convert the data from the standard
456 form to his own internal form.
458 Since 7.15.4 at least line endings are converted.
460 7.6 FTP with NULs in URL parts
462 FTP URLs passed to curl may contain NUL (0x00) in the RFC 1738 <user>,
463 <password>, and <fpath> components, encoded as "%00". The problem is that
464 curl_unescape does not detect this, but instead returns a shortened C string.
465 From a strict FTP protocol standpoint, NUL is a valid character within RFC
466 959 <string>, so the way to handle this correctly in curl would be to use a
467 data structure other than a plain C string, one that can handle embedded NUL
468 characters. From a practical standpoint, most FTP servers would not
469 meaningfully support NUL characters within RFC 959 <string>, anyway (e.g.,
470 Unix pathnames may not contain NUL).
472 7.7 FTP and empty path parts in the URL
474 libcurl ignores empty path parts in FTP URLs, whereas RFC1738 states that
475 such parts should be sent to the server as 'CWD ' (without an argument). The
476 only exception to this rule, is that we knowingly break this if the empty
477 part is first in the path, as then we use the double slashes to indicate that
478 the user wants to reach the root dir (this exception SHALL remain even when
481 7.8 Premature transfer end but healthy control channel
483 When 'multi_done' is called before the transfer has been completed the normal
484 way, it is considered a "premature" transfer end. In this situation, libcurl
485 closes the connection assuming it doesn't know the state of the connection so
486 it can't be reused for subsequent requests.
488 With FTP however, this isn't necessarily true but there are a bunch of
489 situations (listed in the ftp_done code) where it *could* keep the connection
490 alive even in this situation - but the current code doesn't. Fixing this would
491 allow libcurl to reuse FTP connections better.
495 8.1 TELNET and time limtiations don't work
497 When using telnet, the time limitation options don't work.
498 https://curl.haxx.se/bug/view.cgi?id=846
500 8.2 Microsoft telnet server
502 There seems to be a problem when connecting to the Microsoft telnet server.
503 https://curl.haxx.se/bug/view.cgi?id=649
508 9.1 SFTP doesn't do CURLOPT_POSTQUOTE correct
510 When libcurl sends CURLOPT_POSTQUOTE commands when connected to a SFTP server
511 using the multi interface, the commands are not being sent correctly and
512 instead the connection is "cancelled" (the operation is considered done)
513 prematurely. There is a half-baked (busy-looping) patch provided in the bug
514 report but it cannot be accepted as-is. See
515 https://curl.haxx.se/bug/view.cgi?id=748
520 10.1 SOCKS proxy connections are done blocking
522 Both SOCKS5 and SOCKS4 proxy connections are done blocking, which is very bad
523 when used with the multi interface.
525 10.2 SOCKS don't support timeouts
527 The SOCKS4 connection codes don't properly acknowledge (connect) timeouts.
528 According to bug #1556528, even the SOCKS5 connect code does not do it right:
529 https://curl.haxx.se/bug/view.cgi?id=604
531 When connecting to a SOCK proxy, the (connect) timeout is not properly
532 acknowledged after the actual TCP connect (during the SOCKS "negotiate"
537 libcurl doesn't support FTPS over a SOCKS proxy.
539 10.4 active FTP over a SOCKS
541 libcurl doesn't support active FTP over a SOCKS proxy
546 11.1 Curl leaks .onion hostnames in DNS
548 Curl sends DNS requests for hostnames with a .onion TLD. This leaks
549 information about what the user is attempting to access, and violates this
550 requirement of RFC7686: https://tools.ietf.org/html/rfc7686
552 Issue: https://github.com/curl/curl/issues/543
554 11.2 error buffer not set if connection to multiple addresses fails
556 If you ask libcurl to resolve a hostname like example.com to IPv6 addresses
557 only. But you only have IPv4 connectivity. libcurl will correctly fail with
558 CURLE_COULDNT_CONNECT. But the error buffer set by CURLOPT_ERRORBUFFER
559 remains empty. Issue: https://github.com/curl/curl/issues/544
561 11.3 c-ares deviates from stock resolver on http://1346569778
563 When using the socket resolvers, that URL becomes:
565 * Rebuilt URL to: http://1346569778/
566 * Trying 80.67.6.50...
568 but with c-ares it instead says "Could not resolve: 1346569778 (Domain name
571 See https://github.com/curl/curl/issues/893
574 12. LDAP and OpenLDAP
576 12.1 OpenLDAP hangs after returning results
578 By configuration defaults, openldap automatically chase referrals on
579 secondary socket descriptors. The OpenLDAP backend is asynchronous and thus
580 should monitor all socket descriptors involved. Currently, these secondary
581 descriptors are not monitored, causing openldap library to never receive
584 As a temporary workaround, disable referrals chasing by configuration.
586 The fix is not easy: proper automatic referrals chasing requires a
587 synchronous bind callback and monitoring an arbitrary number of socket
588 descriptors for a single easy handle (currently limited to 5).
590 Generic LDAP is synchronous: OK.
592 See https://github.com/curl/curl/issues/622 and
593 https://curl.haxx.se/mail/lib-2016-01/0101.html
598 13.1 --interface for ipv6 binds to unusable IP address
600 Since IPv6 provides a lot of addresses with different scope, binding to an
601 IPv6 address needs to take the proper care so that it doesn't bind to a
602 locally scoped address as that is bound to fail.
604 https://github.com/curl/curl/issues/686