5 \___|\___/|_| \_\_____|
7 Things that could be nice to do in the future
9 Things to do in project cURL. Please tell us what you think, contribute and
10 send us patches that improve things!
12 All bugs documented in the KNOWN_BUGS document are subject for fixing!
17 1.4 signal-based resolver timeouts
18 1.5 get rid of PATH_MAX
19 1.6 progress callback without doubles
20 1.7 Happy Eyeball dual stack connect
22 2. libcurl - multi interface
24 2.2 Remove easy interface internally
25 2.4 Fix HTTP Pipelining for PUT
32 4.2 Alter passive/active on failure and retry
33 4.3 Earlier bad letter detection
34 4.4 REST for large files
39 5.1 Better persistency for HTTP 1.0
40 5.2 support FF3 sqlite cookie files
41 5.3 Rearrange request header order
45 6.2 ditch telnet-specific select
46 6.3 feature negotiation debug data
47 6.4 send data in chunks
50 7.1 Disable specific versions
51 7.2 Provide mutex locking API
52 7.3 Evaluate SSL patches
53 7.4 Cache OpenSSL contexts
54 7.5 Export session ids
55 7.6 Provide callback for cert verification
56 7.7 Support other SSL libraries
57 7.9 improve configure --with-ssl
65 9.1 Other authentication mechanisms
66 9.2 Specify the preferred authentication mechanism
71 10.1 APOP Authentication
72 10.2 Other authentication mechanisms
83 13.3 prevent file overwriting
84 13.4 simultaneous parallel transfers
85 13.5 provide formpost headers
86 13.6 url-specific options
88 13.8 warning when setting an option
89 13.9 IPv6 addresses with globbing
96 15.2 nicer lacking perl message
97 15.3 more protocols supported
98 15.4 more platforms supported
101 16.1 http-style HEAD output for ftp
102 16.2 combine error codes
103 16.3 extend CURLOPT_SOCKOPTFUNCTION prototype
105 17. Next major release
106 17.1 cleanup return codes
107 17.2 remove obsolete defines
109 17.4 remove several functions
110 17.5 remove CURLOPT_FAILONERROR
111 17.6 remove CURLOPT_DNS_USE_GLOBAL_CACHE
112 17.7 remove progress meter from libcurl
113 17.8 remove 'curl_httppost' from public
114 17.9 have form functions use CURL handle argument
116 ==============================================================================
120 1.2 More data sharing
122 curl_share_* functions already exist and work, and they can be extended to
123 share more. For example, enable sharing of the ares channel and the
128 Use 'struct lifreq' and SIOCGLIFADDR instead of 'struct ifreq' and
129 SIOCGIFADDR on newer Solaris versions as they claim the latter is obsolete.
130 To support ipv6 interface addresses for network interfaces properly.
132 1.4 signal-based resolver timeouts
134 libcurl built without an asynchronous resolver library uses alarm() to time
135 out DNS lookups. When a timeout occurs, this causes libcurl to jump from the
136 signal handler back into the library with a sigsetjmp, which effectively
137 causes libcurl to continue running within the signal handler. This is
138 non-portable and could cause problems on some platforms. A discussion on the
139 problem is available at http://curl.haxx.se/mail/lib-2008-09/0197.html
141 Also, alarm() provides timeout resolution only to the nearest second. alarm
142 ought to be replaced by setitimer on systems that support it.
144 1.5 get rid of PATH_MAX
146 Having code use and rely on PATH_MAX is not nice:
147 http://insanecoding.blogspot.com/2007/11/pathmax-simply-isnt.html
149 Currently the SSH based code uses it a bit, but to remove PATH_MAX from there
150 we need libssh2 to properly tell us when we pass in a too small buffer and
151 its current API (as of libssh2 1.2.7) doesn't.
153 1.6 progress callback without doubles
155 The progress callback was introduced way back in the days and the choice to
156 use doubles in the arguments was possibly good at the time. Today the doubles
157 only confuse users and make the amounts less precise. We should introduce
158 another progress callback option that take precedence over the old one and
159 have both co-exist for a forseeable time until we can remove the double-using
162 1.7 Happy Eyeball dual stack connect
164 In order to make alternative technologies not suffer when transitioning, like
165 when introducing IPv6 as an alternative to IPv4 and there are more than one
166 option existing simultaneously there are reasons to reconsider internal
169 To make libcurl do blazing fast IPv6 in a dual-stack configuration, this needs
172 http://tools.ietf.org/html/rfc6555
175 2. libcurl - multi interface
177 2.1 More non-blocking
179 Make sure we don't ever loop because of non-blocking sockets returning
180 EWOULDBLOCK or similar. Blocking cases include:
182 - Name resolves on non-windows unless c-ares is used
183 - NSS SSL connections
184 - HTTP proxy CONNECT operations
185 - SOCKS proxy handshakes
188 - The "DONE" operation (post transfer protocol-specific actions) for the
189 protocols SFTP, SMTP, FTP. Fixing Curl_done() for this is a worthy task.
191 2.2 Remove easy interface internally
193 Make curl_easy_perform() a wrapper-function that simply creates a multi
194 handle, adds the easy handle to it, runs curl_multi_perform() until the
195 transfer is done, then detach the easy handle, destroy the multi handle and
196 return the easy handle's return code. This will thus make everything
197 internally use and assume the multi interface. The select()-loop should use
200 2.4 Fix HTTP Pipelining for PUT
202 HTTP Pipelining can be a way to greatly enhance performance for multiple
203 serial requests and currently libcurl only supports that for HEAD and GET
204 requests but it should also be possible for PUT.
216 HOST is a suggested command in the works for a client to tell which host name
217 to use, to offer FTP servers named-based virtual hosting:
219 http://tools.ietf.org/html/draft-hethmon-mcmurray-ftp-hosts-11
221 4.2 Alter passive/active on failure and retry
223 When trying to connect passively to a server which only supports active
224 connections, libcurl returns CURLE_FTP_WEIRD_PASV_REPLY and closes the
225 connection. There could be a way to fallback to an active connection (and
226 vice versa). http://curl.haxx.se/bug/feature.cgi?id=1754793
228 4.3 Earlier bad letter detection
230 Make the detection of (bad) %0d and %0a codes in FTP url parts earlier in the
231 process to avoid doing a resolve and connect in vain.
233 4.4 REST for large files
235 REST fix for servers not behaving well on >2GB requests. This should fail if
236 the server doesn't set the pointer to the requested index. The tricky
237 (impossible?) part is to figure out if the server did the right thing or not.
239 4.5 FTP proxy support
241 Support the most common FTP proxies, Philip Newton provided a list allegedly
242 from ncftp. This is not a subject without debate, and is probably not really
243 suitable for libcurl. http://curl.haxx.se/mail/archive-2003-04/0126.html
247 FTP ASCII transfers do not follow RFC959. They don't convert the data
252 5.1 Better persistency for HTTP 1.0
254 "Better" support for persistent connections over HTTP 1.0
255 http://curl.haxx.se/bug/feature.cgi?id=1089001
257 5.2 support FF3 sqlite cookie files
259 Firefox 3 is changing from its former format to a a sqlite database instead.
260 We should consider how (lib)curl can/should support this.
261 http://curl.haxx.se/bug/feature.cgi?id=1871388
263 5.3 Rearrange request header order
265 Server implementors often make an effort to detect browser and to reject
266 clients it can detect to not match. One of the last details we cannot yet
267 control in libcurl's HTTP requests, which also can be exploited to detect
268 that libcurl is in fact used even when it tries to impersonate a browser, is
269 the order of the request headers. I propose that we introduce a new option in
270 which you give headers a value, and then when the HTTP request is built it
271 sorts the headers based on that number. We could then have internally created
272 headers use a default value so only headers that need to be moved have to be
280 Reading input (to send to the remote server) on stdin is a crappy solution for
281 library purposes. We need to invent a good way for the application to be able
282 to provide the data to send.
284 6.2 ditch telnet-specific select
286 Move the telnet support's network select() loop go away and merge the code
287 into the main transfer loop. Until this is done, the multi interface won't
290 6.3 feature negotiation debug data
292 Add telnet feature negotiation data to the debug callback as header data.
294 6.4 send data in chunks
296 Currently, telnet sends data one byte at a time. This is fine for interactive
297 use, but inefficient for any other. Sent data should be sent in larger
302 7.1 Disable specific versions
304 Provide an option that allows for disabling specific SSL versions, such as
305 SSLv2 http://curl.haxx.se/bug/feature.cgi?id=1767276
307 7.2 Provide mutex locking API
309 Provide a libcurl API for setting mutex callbacks in the underlying SSL
310 library, so that the same application code can use mutex-locking
311 independently of OpenSSL or GnutTLS being used.
313 7.3 Evaluate SSL patches
315 Evaluate/apply Gertjan van Wingerde's SSL patches:
316 http://curl.haxx.se/mail/lib-2004-03/0087.html
318 7.4 Cache OpenSSL contexts
320 "Look at SSL cafile - quick traces look to me like these are done on every
321 request as well, when they should only be necessary once per ssl context (or
322 once per handle)". The major improvement we can rather easily do is to make
323 sure we don't create and kill a new SSL "context" for every request, but
324 instead make one for every connection and re-use that SSL context in the same
325 style connections are re-used. It will make us use slightly more memory but
326 it will libcurl do less creations and deletions of SSL contexts.
328 7.5 Export session ids
330 Add an interface to libcurl that enables "session IDs" to get
331 exported/imported. Cris Bailiff said: "OpenSSL has functions which can
332 serialise the current SSL state to a buffer of your choice, and recover/reset
333 the state from such a buffer at a later date - this is used by mod_ssl for
334 apache to implement and SSL session ID cache".
336 7.6 Provide callback for cert verification
338 OpenSSL supports a callback for customised verification of the peer
339 certificate, but this doesn't seem to be exposed in the libcurl APIs. Could
340 it be? There's so much that could be done if it were!
342 7.7 Support other SSL libraries
344 Make curl's SSL layer capable of using other free SSL libraries. Such as
345 MatrixSSL (http://www.matrixssl.org/).
347 7.9 improve configure --with-ssl
349 make the configure --with-ssl option first check for OpenSSL, then GnuTLS,
356 Is this even possible?
360 Add a way to check if the connection seems to be alive, to correspond to the
361 SSL_peak() way we use with OpenSSL.
365 libcurl assumes that there are gcrypt functions available when
368 GnuTLS can be built to use libnettle instead as crypto library,
369 which breaks the previously mentioned assumption
371 The correct fix would be to detect which crypto layer that is in use and
372 adapt our code to use that instead of blindly assuming gcrypt.
376 9.1 Other authentication mechanisms
378 Add support for gssapi.
380 9.2 Specify the preferred authentication mechanism
382 Add the ability to specify the preferred authentication mechanism or a list
383 of mechanisms that should be used. Not only that, but the order that is
384 returned by the server during the EHLO response should be honored by curl.
388 Add the ability for the user to specify whether the initial response is
389 included in the AUTH command. Some email servers, such as Microsoft
390 Exchange, can work with either whilst others need to have the initial
391 response sent separately:
393 http://curl.haxx.se/mail/lib-2012-03/0114.html
397 Add support for pipelining emails.
401 10.1 APOP Authentication
403 Add support for the APOP command rather than using plain text authentication
404 (USER and PASS) as this is very week security wise. Note: The APOP command
405 is specified as "APOP <username> <md5 password>", however, it isn't
406 supported by all mail servers.
408 10.2 Other authentication mechanisms
410 SASL offers support for additional authentication mechanisms via the AUTH
411 command. Detection of an email server's support for SASL authentication
412 can be detected via the CAPA command whilst a list of supported mechanisms
413 can be retrieved with an empty AUTH command.
417 Being able to specify the preferred authentication mechanim in the URL as
418 per RFC-2384 (http://tools.ietf.org/html/rfc2384).
426 There's no RFC for the protocol or an URI/URL format. An implementation
427 should most probably use an existing rsync library, such as librsync.
433 "curl --sync http://example.com/feed[1-100].rss" or
434 "curl --sync http://example.net/{index,calendar,history}.html"
436 Downloads a range or set of URLs using the remote name, but only if the
437 remote file is newer than the local file. A Last-Modified HTTP date header
438 should also be used to set the mod date on the downloaded file.
442 Globbing support for -d and -F, as in 'curl -d "name=foo[0-9]" URL'.
443 This is easily scripted though.
445 13.3 prevent file overwriting
447 Add an option that prevents cURL from overwriting existing local files. When
448 used, and there already is an existing file with the target file name
449 (either -O or -o), a number should be appended (and increased if already
450 existing). So that index.html becomes first index.html.1 and then
453 13.4 simultaneous parallel transfers
455 The client could be told to use maximum N simultaneous parallel transfers and
456 then just make sure that happens. It should of course not make more than one
457 connection to the same remote host. This would require the client to use the
458 multi interface. http://curl.haxx.se/bug/feature.cgi?id=1558595
460 13.5 provide formpost headers
462 Extending the capabilities of the multipart formposting. How about leaving
463 the ';type=foo' syntax as it is and adding an extra tag (headers) which
464 works like this: curl -F "coolfiles=@fil1.txt;headers=@fil1.hdr" where
465 fil1.hdr contains extra headers like
467 Content-Type: text/plain; charset=KOI8-R"
468 Content-Transfer-Encoding: base64
469 X-User-Comment: Please don't use browser specific HTML code
471 which should overwrite the program reasonable defaults (plain/text,
474 13.6 url-specific options
476 Provide a way to make options bound to a specific URL among several on the
477 command line. Possibly by letting ':' separate options between URLs,
480 curl --data foo --url url.com : \
482 --url url3.com --data foo3
484 (More details: http://curl.haxx.se/mail/archive-2004-07/0133.html)
486 The example would do a POST-GET-POST combination on a single command line.
488 13.7 metalink support
490 Add metalink support to curl (http://www.metalinker.org/). This is most useful
491 with simultaneous parallel transfers (11.6) but not necessary.
493 13.8 warning when setting an option
495 Display a warning when libcurl returns an error when setting an option.
496 This can be useful to tell when support for a particular feature hasn't been
497 compiled into the library.
499 13.9 IPv6 addresses with globbing
501 Currently the command line client needs to get url globbing disabled (with
502 -g) for it to support IPv6 numerical addresses. This is a rather silly flaw
503 that should be corrected. It probably involves a smarter detection of the
510 Consider extending 'roffit' to produce decent ASCII output, and use that
511 instead of (g)nroff when building src/hugehelp.c
517 Make our own version of stunnel for simple port forwarding to enable HTTPS
518 and FTP-SSL tests without the stunnel dependency, and it could allow us to
519 provide test tools built with either OpenSSL or GnuTLS
521 15.2 nicer lacking perl message
523 If perl wasn't found by the configure script, don't attempt to run the tests
524 but explain something nice why it doesn't.
526 15.3 more protocols supported
528 Extend the test suite to include more protocols. The telnet could just do ftp
529 or http operations (for which we have test servers).
531 15.4 more platforms supported
533 Make the test suite work on more platforms. OpenBSD and Mac OS. Remove
534 fork()s and it should become even more portable.
538 16.1 http-style HEAD output for ftp
540 #undef CURL_FTP_HTTPSTYLE_HEAD in lib/ftp.c to remove the HTTP-style headers
541 from being output in NOBODY requests over ftp
543 16.2 combine error codes
545 Combine some of the error codes to remove duplicates. The original
546 numbering should not be changed, and the old identifiers would be
547 macroed to the new ones in an CURL_NO_OLDIES section to help with
548 backward compatibility.
550 Candidates for removal and their replacements:
552 CURLE_FILE_COULDNT_READ_FILE => CURLE_REMOTE_FILE_NOT_FOUND
553 CURLE_FTP_COULDNT_RETR_FILE => CURLE_REMOTE_FILE_NOT_FOUND
554 CURLE_FTP_COULDNT_USE_REST => CURLE_RANGE_ERROR
555 CURLE_FUNCTION_NOT_FOUND => CURLE_FAILED_INIT
556 CURLE_LDAP_INVALID_URL => CURLE_URL_MALFORMAT
557 CURLE_TFTP_NOSUCHUSER => CURLE_TFTP_ILLEGAL
558 CURLE_TFTP_NOTFOUND => CURLE_REMOTE_FILE_NOT_FOUND
559 CURLE_TFTP_PERM => CURLE_REMOTE_ACCESS_DENIED
561 16.3 extend CURLOPT_SOCKOPTFUNCTION prototype
563 The current prototype only provides 'purpose' that tells what the
564 connection/socket is for, but not any protocol or similar. It makes it hard
565 for applications to differentiate on TCP vs UDP and even HTTP vs FTP and
568 17. Next major release
570 17.1 cleanup return codes
572 curl_easy_cleanup() returns void, but curl_multi_cleanup() returns a
573 CURLMcode. These should be changed to be the same.
575 17.2 remove obsolete defines
577 remove obsolete defines from curl/curl.h
581 make several functions use size_t instead of int in their APIs
583 17.4 remove several functions
585 remove the following functions from the public API:
589 curl_mprintf (and variations)
595 They will instead become curlx_ - alternatives. That makes the curl app
596 still capable of using them, by building with them from source.
598 These functions have no purpose anymore:
602 curl_multi_socket_all
604 17.5 remove CURLOPT_FAILONERROR
606 Remove support for CURLOPT_FAILONERROR, it has gotten too kludgy and weird
607 internally. Let the app judge success or not for itself.
609 17.6 remove CURLOPT_DNS_USE_GLOBAL_CACHE
611 Remove support for a global DNS cache. Anything global is silly, and we
612 already offer the share interface for the same functionality but done
615 17.7 remove progress meter from libcurl
617 The internally provided progress meter output doesn't belong in the library.
618 Basically no application wants it (apart from curl) but instead applications
619 can and should do their own progress meters using the progress callback.
621 The progress callback should then be bumped as well to get proper 64bit
622 variable types passed to it instead of doubles so that big files work
625 17.8 remove 'curl_httppost' from public
627 curl_formadd() was made to fill in a public struct, but the fact that the
628 struct is public is never really used by application for their own advantage
629 but instead often restricts how the form functions can or can't be modified.
631 Changing them to return a private handle will benefit the implementation and
632 allow us much greater freedoms while still maintining a solid API and ABI.
634 17.9 have form functions use CURL handle argument
636 curl_formadd() and curl_formget() both currently have no CURL handle
637 argument, but both can use a callback that is set in the easy handle, and
638 thus curl_formget() with callback cannot function without first having
639 curl_easy_perform() (or similar) called - which is hard to grasp and a design