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 33. Doing multi-pass HTTP authentication on a non-default port does not work.
7 This happens because the multi-pass code abuses the redirect following code
8 for doing multiple requests, and when we following redirects to an absolute
9 URL we must use the newly specified port and not the one specified in the
10 original URL. A proper fix to this would need to separate the negotiation
11 "redirect" from an actual redirect.
13 32. (At least on Windows) If libcurl is built with c-ares and there's no DNS
14 server configured in the system, the ares_init() call fails and thus
15 curl_easy_init() fails as well. This causes weird effects for people who use
16 numerical IP addresses only.
18 31. "curl-config --libs" will include details set in LDFLAGS when configure is
19 run that might be needed only for building libcurl. Similarly, it might
20 include options that perhaps aren't suitable both for static and dynamic
21 linking. Further, curl-config --cflags suffers from the same effects with
24 30. You need to use -g to the command line tool in order to use RFC2732-style
25 IPv6 numerical addresses in URLs.
27 29. IPv6 URLs with zone ID is not supported.
28 http://www.ietf.org/internet-drafts/draft-fenner-literal-zone-02.txt
29 specifies the use of a plus sign instead of a percent when specifying zone
30 IDs in URLs to get around the problem of percent signs being
31 special. According to the reporter, Firefox deals with the URL _with_ a
32 percent letter (which seems like a blatant URL spec violation).
34 See http://curl.haxx.se/bug/view.cgi?id=1371118
36 28. The TFTP code is not portable and will fail on some architectures.
38 26. NTLM authentication using SSPI (on Windows) when (lib)curl is running in
39 "system context" will make it use wrong(?) user name - at least when compared
40 to what winhttp does. See http://curl.haxx.se/bug/view.cgi?id=1281867
42 25. When doing a CONNECT request with curl it doesn't properly handle if the
43 proxy closes the connection within the authentication "negotiation phase".
44 Like if you do HTTPS or similar over a proxy and you use perhaps
45 --proxy-anyauth. There's work in progress on this problem, and a recent
46 patch was posted here: http://curl.haxx.se/mail/lib-2005-08/0074.html
48 23. We don't support SOCKS for IPv6. We don't support FTPS over a SOCKS proxy.
49 We don't have any test cases for SOCKS proxy. We probably have even more
50 bugs and lack of features when a SOCKS proxy is used. And there seem to be a
51 problem with SOCKS when doing FTP: See
52 http://curl.haxx.se/bug/view.cgi?id=1371540
54 22. Sending files to a FTP server using curl on VMS, might lead to curl
55 complaining on "unaligned file size" on completion. The problem is related
56 to VMS file structures and the perceived file sizes stat() returns. A
57 possible fix would involve sending a "STRU VMS" command.
58 http://curl.haxx.se/bug/view.cgi?id=1156287
60 21. FTP ASCII transfers do not follow RFC959. They don't convert the data
61 accordingly (not for sending nor for receiving). RFC 959 section 3.1.1.1
62 clearly describes how this should be done:
64 The sender converts the data from an internal character representation to
65 the standard 8-bit NVT-ASCII representation (see the Telnet
66 specification). The receiver will convert the data from the standard
67 form to his own internal form.
69 Since 7.15.4 at least line endings are converted.
71 19. FTP 3rd party transfers with the multi interface doesn't work. Test:
72 define CURL_MULTIEASY, rebuild curl, run test case 230 - 232.
74 16. FTP URLs passed to curl may contain NUL (0x00) in the RFC 1738 <user>,
75 <password>, and <fpath> components, encoded as "%00". The problem is that
76 curl_unescape does not detect this, but instead returns a shortened C
77 string. From a strict FTP protocol standpoint, NUL is a valid character
78 within RFC 959 <string>, so the way to handle this correctly in curl would
79 be to use a data structure other than a plain C string, one that can handle
80 embedded NUL characters. From a practical standpoint, most FTP servers
81 would not meaningfully support NUL characters within RFC 959 <string>,
82 anyway (e.g., UNIX pathnames may not contain NUL).
84 14. Test case 165 might fail on system which has libidn present, but with an
85 old iconv version (2.1.3 is a known bad version), since it doesn't recognize
86 the charset when named ISO8859-1. Changing the name to ISO-8859-1 makes the
87 test pass, but instead makes it fail on Solaris hosts that use its native
90 13. curl version 7.12.2 fails on AIX if compiled with --enable-ares.
91 The workaround is to combine --enable-ares with --disable-shared
93 12. When connecting to a SOCKS proxy, the (connect) timeout is not properly
94 acknowledged after the actual TCP connect (during the SOCKS "negotiate"
95 phase). Pointed out by Lucas. Fix: need to select() and timeout properly.
97 11. Using configure --disable-[protocol] may cause 'make test' to fail for
98 tests using the disabled protocol(s).
100 10. To get HTTP Negotiate authentication to work fine, you need to provide a
101 (fake) user name (this concerns both curl and the lib) because the code
102 wrongly only considers authentication if there's a user name provided.
103 http://curl.haxx.se/bug/view.cgi?id=1004841. How?
104 http://curl.haxx.se/mail/lib-2004-08/0182.html
106 9. --limit-rate using -d or -F does not work. This is because the limit logic
107 is provided by the curl app in its read/write callbacks, and when doing
108 -d/-F the callbacks aren't used! http://curl.haxx.se/bug/view.cgi?id=921395
110 8. Doing resumed upload over HTTP does not work with '-C -', because curl
111 doesn't do a HEAD first to get the initial size. This needs to be done
112 manually for HTTP PUT resume to work, and then '-C [index]'.
114 7. CURLOPT_USERPWD and CURLOPT_PROXYUSERPWD have no way of providing user names
115 that contain a colon. This can't be fixed easily in a backwards compatible
116 way without adding new options (and then, they should most probably allow
117 setting user name and password separately).
119 6. libcurl ignores empty path parts in FTP URLs, whereas RFC1738 states that
120 such parts should be sent to the server as 'CWD ' (without an argument).
121 The only exception to this rule, is that we knowingly break this if the
122 empty part is first in the path, as then we use the double slashes to
123 indicate that the user wants to reach the root dir (this exception SHALL
124 remain even when this bug is fixed).
126 5. libcurl doesn't treat the content-length of compressed data properly, as
127 it seems HTTP servers send the *uncompressed* length in that header and
128 libcurl thinks of it as the *compressed* length. Some explanations are here:
129 http://curl.haxx.se/mail/lib-2003-06/0146.html
131 2. If a HTTP server responds to a HEAD request and includes a body (thus
132 violating the RFC2616), curl won't wait to read the response but just stop
133 reading and return back. If a second request (let's assume a GET) is then
134 immediately made to the same server again, the connection will be re-used
135 fine of course, and the second request will be sent off but when the
136 response is to get read, the previous response-body is what curl will read
137 and havoc is what happens.
138 More details on this is found in this libcurl mailing list thread:
139 http://curl.haxx.se/mail/lib-2002-08/0000.html