1 These are problems known to exist at the time of this release. Feel free to
2 join in and help us correct one or more of these! Also be sure to check the
3 changelog of the current development status, as one or more of these problems
4 may have been fixed since this was written!
6 88. libcurl doesn't support CURLINFO_FILETIME for SFTP transfers and thus
7 curl's -R option also doesn't work then.
9 87. -J/--remote-header-name doesn't decode %-encoded file names. RFC6266
10 details how it should be done. The can of worm is basically that we have no
11 charset handling in curl and ascii >=128 is a challenge for us. Not to
12 mention that decoding also means that we need to check for nastiness that is
13 attempted, like "../" sequences and the like. Probably everything to the left
14 of any embedded slashes should be cut off.
15 http://curl.haxx.se/bug/view.cgi?id=1294
17 86. The disconnect commands (LOGOUT and QUIT) may not be sent by IMAP, POP3
18 and SMTP if a failure occurs during the authentication phase of a
21 85. Wrong STARTTRANSFER timer accounting for POST requests
22 Timer works fine with GET requests, but while using POST the time for
23 CURLINFO_STARTTRANSFER_TIME is wrong. While using POST
24 CURLINFO_STARTTRANSFER_TIME minus CURLINFO_PRETRANSFER_TIME is near to zero
26 http://curl.haxx.se/bug/view.cgi?id=1213
28 84. CURLINFO_SSL_VERIFYRESULT is only implemented for the OpenSSL and NSS
29 backends, so relying on this information in a generic app is flaky.
31 82. When building with the Windows Borland compiler, it fails because the
32 "tlib" tool doesn't support hyphens (minus signs) in file names and we have
34 http://curl.haxx.se/bug/view.cgi?id=1222
36 81. When using -J (with -O), automatically resumed downloading together with
37 "-C -" fails. Without -J the same command line works! This happens because
38 the resume logic is worked out before the target file name (and thus its
39 pre-transfer size) has been figured out!
40 http://curl.haxx.se/bug/view.cgi?id=1169
42 80. Curl doesn't recognize certificates in DER format in keychain, but it
44 http://curl.haxx.se/bug/view.cgi?id=1065
46 79. SMTP. When sending data to multiple recipients, curl will abort and return
47 failure if one of the recipients indicate failure (on the "RCPT TO"
48 command). Ordinary mail programs would proceed and still send to the ones
49 that can receive data. This is subject for change in the future.
50 http://curl.haxx.se/bug/view.cgi?id=1116
52 78. curl and libcurl don't always signal the client properly when "sending"
53 zero bytes files - it makes for example the command line client not creating
54 any file at all. Like when using FTP.
55 http://curl.haxx.se/bug/view.cgi?id=1063
57 76. The SOCKET type in Win64 is 64 bits large (and thus so is curl_socket_t on
58 that platform), and long is only 32 bits. It makes it impossible for
59 curl_easy_getinfo() to return a socket properly with the CURLINFO_LASTSOCKET
60 option as for all other operating systems.
62 75. NTLM authentication involving unicode user name or password only works
63 properly if built with UNICODE defined together with the WinSSL/schannel
64 backend. The original problem was mentioned in:
65 http://curl.haxx.se/mail/lib-2009-10/0024.html
66 http://curl.haxx.se/bug/view.cgi?id=896
68 The WinSSL/schannel version verified to work as mentioned in
69 http://curl.haxx.se/mail/lib-2012-07/0073.html
71 73. if a connection is made to a FTP server but the server then just never
72 sends the 220 response or otherwise is dead slow, libcurl will not
73 acknowledge the connection timeout during that phase but only the "real"
74 timeout - which may surprise users as it is probably considered to be the
75 connect phase to most people. Brought up (and is being misunderstood) in:
76 http://curl.haxx.se/bug/view.cgi?id=856
78 72. "Pausing pipeline problems."
79 http://curl.haxx.se/mail/lib-2009-07/0214.html
81 70. Problem re-using easy handle after call to curl_multi_remove_handle
82 http://curl.haxx.se/mail/lib-2009-07/0249.html
84 68. "More questions about ares behavior".
85 http://curl.haxx.se/mail/lib-2009-08/0012.html
87 67. When creating multipart formposts. The file name part can be encoded with
88 something beyond ascii but currently libcurl will only pass in the verbatim
89 string the app provides. There are several browsers that already do this
90 encoding. The key seems to be the updated draft to RFC2231:
91 http://tools.ietf.org/html/draft-reschke-rfc2231-in-http-02
93 66. When using telnet, the time limitation options don't work.
94 http://curl.haxx.se/bug/view.cgi?id=846
96 65. When doing FTP over a socks proxy or CONNECT through HTTP proxy and the
97 multi interface is used, libcurl will fail if the (passive) TCP connection
98 for the data transfer isn't more or less instant as the code does not
99 properly wait for the connect to be confirmed. See test case 564 for a first
102 63. When CURLOPT_CONNECT_ONLY is used, the handle cannot reliably be re-used
103 for any further requests or transfers. The work-around is then to close that
104 handle with curl_easy_cleanup() and create a new. Some more details:
105 http://curl.haxx.se/mail/lib-2009-04/0300.html
107 61. If an upload using Expect: 100-continue receives an HTTP 417 response,
108 it ought to be automatically resent without the Expect:. A workaround is
109 for the client application to redo the transfer after disabling Expect:.
110 http://curl.haxx.se/mail/archive-2008-02/0043.html
112 60. libcurl closes the connection if an HTTP 401 reply is received while it
113 is waiting for the the 100-continue response.
114 http://curl.haxx.se/mail/lib-2008-08/0462.html
116 58. It seems sensible to be able to use CURLOPT_NOBODY and
117 CURLOPT_FAILONERROR with FTP to detect if a file exists or not, but it is
118 not working: http://curl.haxx.se/mail/lib-2008-07/0295.html
120 56. When libcurl sends CURLOPT_POSTQUOTE commands when connected to a SFTP
121 server using the multi interface, the commands are not being sent correctly
122 and instead the connection is "cancelled" (the operation is considered done)
123 prematurely. There is a half-baked (busy-looping) patch provided in the bug
124 report but it cannot be accepted as-is. See
125 http://curl.haxx.se/bug/view.cgi?id=748
127 55. libcurl fails to build with MIT Kerberos for Windows (KfW) due to KfW's
128 library header files exporting symbols/macros that should be kept private
129 to the KfW library. See ticket #5601 at http://krbdev.mit.edu/rt/
131 52. Gautam Kachroo's issue that identifies a problem with the multi interface
132 where a connection can be re-used without actually being properly
134 http://curl.haxx.se/mail/lib-2008-01/0277.html
136 49. If using --retry and the transfer timeouts (possibly due to using -m or
137 -y/-Y) the next attempt doesn't resume the transfer properly from what was
138 downloaded in the previous attempt but will truncate and restart at the
139 original position where it was at before the previous failed attempt. See
140 http://curl.haxx.se/mail/lib-2008-01/0080.html and Mandriva bug report
141 https://qa.mandriva.com/show_bug.cgi?id=22565
143 48. If a CONNECT response-headers are larger than BUFSIZE (16KB) when the
144 connection is meant to be kept alive (like for NTLM proxy auth), the
145 function will return prematurely and will confuse the rest of the HTTP
146 protocol code. This should be very rare.
148 43. There seems to be a problem when connecting to the Microsoft telnet server.
149 http://curl.haxx.se/bug/view.cgi?id=649
151 41. When doing an operation over FTP that requires the ACCT command (but not
152 when logging in), the operation will fail since libcurl doesn't detect this
153 and thus fails to issue the correct command:
154 http://curl.haxx.se/bug/view.cgi?id=635
156 39. Steffen Rumler's Race Condition in Curl_proxyCONNECT:
157 http://curl.haxx.se/mail/lib-2007-01/0045.html
159 38. Kumar Swamy Bhatt's problem in ftp/ssl "LIST" operation:
160 http://curl.haxx.se/mail/lib-2007-01/0103.html
162 35. Both SOCKS5 and SOCKS4 proxy connections are done blocking, which is very
163 bad when used with the multi interface.
165 34. The SOCKS4 connection codes don't properly acknowledge (connect) timeouts.
166 Also see #12. According to bug #1556528, even the SOCKS5 connect code does
167 not do it right: http://curl.haxx.se/bug/view.cgi?id=604
169 31. "curl-config --libs" will include details set in LDFLAGS when configure is
170 run that might be needed only for building libcurl. Further, curl-config
171 --cflags suffers from the same effects with CFLAGS/CPPFLAGS.
173 26. NTLM authentication using SSPI (on Windows) when (lib)curl is running in
174 "system context" will make it use wrong(?) user name - at least when compared
175 to what winhttp does. See http://curl.haxx.se/bug/view.cgi?id=535
177 23. SOCKS-related problems:
178 B) libcurl doesn't support FTPS over a SOCKS proxy.
179 E) libcurl doesn't support active FTP over a SOCKS proxy
181 We probably have even more bugs and lack of features when a SOCKS proxy is
184 21. FTP ASCII transfers do not follow RFC959. They don't convert the data
185 accordingly (not for sending nor for receiving). RFC 959 section 3.1.1.1
186 clearly describes how this should be done:
188 The sender converts the data from an internal character representation to
189 the standard 8-bit NVT-ASCII representation (see the Telnet
190 specification). The receiver will convert the data from the standard
191 form to his own internal form.
193 Since 7.15.4 at least line endings are converted.
195 16. FTP URLs passed to curl may contain NUL (0x00) in the RFC 1738 <user>,
196 <password>, and <fpath> components, encoded as "%00". The problem is that
197 curl_unescape does not detect this, but instead returns a shortened C
198 string. From a strict FTP protocol standpoint, NUL is a valid character
199 within RFC 959 <string>, so the way to handle this correctly in curl would
200 be to use a data structure other than a plain C string, one that can handle
201 embedded NUL characters. From a practical standpoint, most FTP servers
202 would not meaningfully support NUL characters within RFC 959 <string>,
203 anyway (e.g., Unix pathnames may not contain NUL).
205 14. Test case 165 might fail on a system which has libidn present, but with an
206 old iconv version (2.1.3 is a known bad version), since it doesn't recognize
207 the charset when named ISO8859-1. Changing the name to ISO-8859-1 makes the
208 test pass, but instead makes it fail on Solaris hosts that use its native
211 13. curl version 7.12.2 fails on AIX if compiled with --enable-ares.
212 The workaround is to combine --enable-ares with --disable-shared
214 12. When connecting to a SOCKS proxy, the (connect) timeout is not properly
215 acknowledged after the actual TCP connect (during the SOCKS "negotiate"
218 10. To get HTTP Negotiate (SPNEGO) authentication to work fine, you need to
219 provide a (fake) user name (this concerns both curl and the lib) because the
220 code wrongly only considers authentication if there's a user name provided.
221 http://curl.haxx.se/bug/view.cgi?id=440 How?
222 http://curl.haxx.se/mail/lib-2004-08/0182.html
224 8. Doing resumed upload over HTTP does not work with '-C -', because curl
225 doesn't do a HEAD first to get the initial size. This needs to be done
226 manually for HTTP PUT resume to work, and then '-C [index]'.
228 6. libcurl ignores empty path parts in FTP URLs, whereas RFC1738 states that
229 such parts should be sent to the server as 'CWD ' (without an argument).
230 The only exception to this rule, is that we knowingly break this if the
231 empty part is first in the path, as then we use the double slashes to
232 indicate that the user wants to reach the root dir (this exception SHALL
233 remain even when this bug is fixed).
235 5. libcurl doesn't treat the content-length of compressed data properly, as
236 it seems HTTP servers send the *uncompressed* length in that header and
237 libcurl thinks of it as the *compressed* length. Some explanations are here:
238 http://curl.haxx.se/mail/lib-2003-06/0146.html
240 2. If a HTTP server responds to a HEAD request and includes a body (thus
241 violating the RFC2616), curl won't wait to read the response but just stop
242 reading and return back. If a second request (let's assume a GET) is then
243 immediately made to the same server again, the connection will be re-used
244 fine of course, and the second request will be sent off but when the
245 response is to get read, the previous response-body is what curl will read
246 and havoc is what happens.
247 More details on this is found in this libcurl mailing list thread:
248 http://curl.haxx.se/mail/lib-2002-08/0000.html