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 Be aware that these are things that we could do, or have once been considered
13 things we could do. If you want to work on any of these areas, please
14 consider bringing it up for discussions first on the mailing list so that we
15 all agree it is still a good idea for the project!
17 All bugs documented in the KNOWN_BUGS document are subject for fixing!
22 1.4 signal-based resolver timeouts
23 1.5 get rid of PATH_MAX
24 1.6 Modified buffer size approach
25 1.7 Detect when called from within callbacks
26 1.8 CURLOPT_RESOLVE for any port number
27 1.9 Cache negative name resolves
28 1.10 auto-detect proxy
29 1.11 minimize dependencies with dynamically loaded modules
30 1.14 Typesafe curl_easy_setopt()
31 1.15 Monitor connections in the connection pool
32 1.16 Try to URL encode given URL
33 1.17 Add support for IRIs
34 1.18 try next proxy if one doesn't work
35 1.19 Timeout idle connections from the pool
36 1.20 SRV and URI DNS records
37 1.21 API for URL parsing/splitting
38 1.23 Offer API to flush the connection pool
39 1.24 TCP Fast Open for windows
41 2. libcurl - multi interface
43 2.2 Better support for same name resolves
44 2.3 Non-blocking curl_multi_remove_handle()
45 2.4 Split connect and authentication process
46 2.5 Edge-triggered sockets should work
49 3.2 Provide cmake config-file
53 4.2 Alter passive/active on failure and retry
54 4.3 Earlier bad letter detection
55 4.4 REST for large files
57 4.6 GSSAPI via Windows SSPI
58 4.7 STAT for LIST without data connection
61 5.1 Better persistency for HTTP 1.0
62 5.2 support FF3 sqlite cookie files
63 5.3 Rearrange request header order
64 5.4 HTTP Digest using SHA-256
66 5.6 Refuse "downgrade" redirects
67 5.7 Brotli compression
69 5.10 Leave secure cookies alone
73 6.2 ditch telnet-specific select
74 6.3 feature negotiation debug data
78 7.2 Enhanced capability support
79 7.3 Add CURLOPT_MAIL_CLIENT option
83 8.2 Enhanced capability support
86 9.1 Enhanced capability support
89 10.1 SASL based authentication mechanisms
92 11.1 File listing support
93 11.2 Honor file timestamps
95 11.4 Create remote directories
101 13.1 Disable specific versions
102 13.2 Provide mutex locking API
103 13.3 Evaluate SSL patches
104 13.4 Cache/share OpenSSL contexts
105 13.5 Export session ids
106 13.6 Provide callback for cert verification
107 13.7 improve configure --with-ssl
109 13.10 Support SSLKEYLOGFILE
110 13.11 Support intermediate & root pinning for PINNEDPUBLICKEY
115 14.1 SSL engine stuff
116 14.2 check connection
119 15.1 Add support for client certificate authentication
120 15.2 Add support for custom server certificate validation
121 15.3 Add support for the --ciphers option
124 16.1 Other authentication mechanisms
125 16.2 Add QOP support to GSSAPI authentication
126 16.3 Support binary messages (i.e.: non-base64)
130 17.2 SFTP performance
131 17.3 Support better than MD5 hostkey hash
132 17.4 Support CURLOPT_PREQUOTE
134 18. Command line tool
137 18.3 prevent file overwriting
138 18.4 simultaneous parallel transfers
139 18.6 warning when setting an option
140 18.8 offer color-coded HTTP header output
141 18.9 Choose the name of file in braces for complex URLs
142 18.10 improve how curl works in a windows console window
143 18.11 -w output to stderr
144 18.12 keep running, read instructions from pipe/socket
145 18.13 support metalink in http headers
146 18.14 --fail without --location should treat 3xx as a failure
147 18.15 --retry should resume
148 18.16 send only part of --data
149 18.17 consider file name from the redirected URL with -O ?
153 19.2 Enable PIE and RELRO by default
157 20.2 nicer lacking perl message
158 20.3 more protocols supported
159 20.4 more platforms supported
160 20.5 Add support for concurrent connections
161 20.6 Use the RFC6265 test suite
164 21.1 http-style HEAD output for FTP
165 21.2 combine error codes
166 21.3 extend CURLOPT_SOCKOPTFUNCTION prototype
168 22. Next major release
169 22.1 cleanup return codes
170 22.2 remove obsolete defines
172 22.4 remove several functions
173 22.5 remove CURLOPT_FAILONERROR
174 22.6 remove CURLOPT_DNS_USE_GLOBAL_CACHE
175 22.7 remove progress meter from libcurl
176 22.8 remove 'curl_httppost' from public
178 ==============================================================================
182 1.2 More data sharing
184 curl_share_* functions already exist and work, and they can be extended to
185 share more. For example, enable sharing of the ares channel and the
190 Use 'struct lifreq' and SIOCGLIFADDR instead of 'struct ifreq' and
191 SIOCGIFADDR on newer Solaris versions as they claim the latter is obsolete.
192 To support IPv6 interface addresses for network interfaces properly.
194 1.4 signal-based resolver timeouts
196 libcurl built without an asynchronous resolver library uses alarm() to time
197 out DNS lookups. When a timeout occurs, this causes libcurl to jump from the
198 signal handler back into the library with a sigsetjmp, which effectively
199 causes libcurl to continue running within the signal handler. This is
200 non-portable and could cause problems on some platforms. A discussion on the
201 problem is available at https://curl.haxx.se/mail/lib-2008-09/0197.html
203 Also, alarm() provides timeout resolution only to the nearest second. alarm
204 ought to be replaced by setitimer on systems that support it.
206 1.5 get rid of PATH_MAX
208 Having code use and rely on PATH_MAX is not nice:
209 https://insanecoding.blogspot.com/2007/11/pathmax-simply-isnt.html
211 Currently the SSH based code uses it a bit, but to remove PATH_MAX from there
212 we need libssh2 to properly tell us when we pass in a too small buffer and
213 its current API (as of libssh2 1.2.7) doesn't.
215 1.6 Modified buffer size approach
217 Current libcurl allocates a fixed 16K size buffer for download and an
218 additional 16K for upload. They are always unconditionally part of the easy
219 handle. If CRLF translations are requested, an additional 32K "scratch
220 buffer" is allocated. A total of 64K transfer buffers in the worst case.
222 First, while the handles are not actually in use these buffers could be freed
223 so that lingering handles just kept in queues or whatever waste less memory.
225 Secondly, SFTP is a protocol that needs to handle many ~30K blocks at once
226 since each need to be individually acked and therefore libssh2 must be
227 allowed to send (or receive) many separate ones in parallel to achieve high
228 transfer speeds. A current libcurl build with a 16K buffer makes that
229 impossible, but one with a 512K buffer will reach MUCH faster transfers. But
230 allocating 512K unconditionally for all buffers just in case they would like
231 to do fast SFTP transfers at some point is not a good solution either.
233 Dynamically allocate buffer size depending on protocol in use in combination
234 with freeing it after each individual transfer? Other suggestions?
236 1.7 Detect when called from within callbacks
238 We should set a state variable before calling callbacks, so that we
239 subsequently can add code within libcurl that returns error if called within
240 callbacks for when that's not supported.
242 1.8 CURLOPT_RESOLVE for any port number
244 This option allows applications to set a replacement IP address for a given
245 host + port pair. Consider making support for providing a replacement address
246 for the host name on all port numbers.
248 See https://github.com/curl/curl/issues/1264
250 1.9 Cache negative name resolves
252 A name resolve that has failed is likely to fail when made again within a
253 short period of time. Currently we only cache positive responses.
255 1.10 auto-detect proxy
257 libcurl could be made to detect the system proxy setup automatically and use
258 that. On Windows, macOS and Linux desktops for example.
260 The pull-request to use libproxy for this was deferred due to doubts on the
261 reliability of the dependency and how to use it:
262 https://github.com/curl/curl/pull/977
264 libdetectproxy is a (C++) library for detecting the proxy on Windows
265 https://github.com/paulharris/libdetectproxy
267 1.11 minimize dependencies with dynamically loaded modules
269 We can create a system with loadable modules/plug-ins, where these modules
270 would be the ones that link to 3rd party libs. That would allow us to avoid
271 having to load ALL dependencies since only the necessary ones for this
272 app/invoke/used protocols would be necessary to load. See
273 https://github.com/curl/curl/issues/349
275 1.14 Typesafe curl_easy_setopt()
277 One of the most common problems in libcurl using applications is the lack of
278 type checks for curl_easy_setopt() which happens because it accepts varargs
279 and thus can take any type.
281 One possible solution to this is to introduce a few different versions of the
282 setopt version for the different kinds of data you can set.
284 curl_easy_set_num() - sets a long value
286 curl_easy_set_large() - sets a curl_off_t value
288 curl_easy_set_ptr() - sets a pointer
290 curl_easy_set_cb() - sets a callback PLUS its callback data
292 1.15 Monitor connections in the connection pool
294 libcurl's connection cache or pool holds a number of open connections for the
295 purpose of possible subsequent connection reuse. It may contain a few up to a
296 significant amount of connections. Currently, libcurl leaves all connections
297 as they are and first when a connection is iterated over for matching or
298 reuse purpose it is verified that it is still alive.
300 Those connections may get closed by the server side for idleness or they may
301 get a HTTP/2 ping from the peer to verify that they're still alive. By adding
302 monitoring of the connections while in the pool, libcurl can detect dead
303 connections (and close them) better and earlier, and it can handle HTTP/2
304 pings to keep such ones alive even when not actively doing transfers on them.
306 1.16 Try to URL encode given URL
308 Given a URL that for example contains spaces, libcurl could have an option
309 that would try somewhat harder than it does now and convert spaces to %20 and
310 perhaps URL encoded byte values over 128 etc (basically do what the redirect
311 following code already does).
313 https://github.com/curl/curl/issues/514
315 1.17 Add support for IRIs
317 IRIs (RFC 3987) allow localized, non-ascii, names in the URL. To properly
318 support this, curl/libcurl would need to translate/encode the given input
319 from the input string encoding into percent encoded output "over the wire".
321 To make that work smoothly for curl users even on Windows, curl would
322 probably need to be able to convert from several input encodings.
324 1.18 try next proxy if one doesn't work
326 Allow an application to specify a list of proxies to try, and failing to
327 connect to the first go on and try the next instead until the list is
328 exhausted. Browsers support this feature at least when they specify proxies
331 https://github.com/curl/curl/issues/896
333 1.19 Timeout idle connections from the pool
335 libcurl currently keeps connections in its connection pool for an indefinite
336 period of time, until it either gets reused, gets noticed that it has been
337 closed by the server or gets pruned to make room for a new connection.
339 To reduce overhead (especially for when we add monitoring of the connections
340 in the pool), we should introduce a timeout so that connections that have
341 been idle for N seconds get closed.
343 1.20 SRV and URI DNS records
345 Offer support for resolving SRV and URI DNS records for libcurl to know which
346 server to connect to for various protocols (including HTTP!).
348 1.21 API for URL parsing/splitting
350 libcurl has always parsed URLs internally and never exposed any API or
351 features to allow applications to do it. Still most or many applications
352 using libcurl need that ability. In polls to users, we've learned that many
353 libcurl users would like to see and use such an API.
355 1.23 Offer API to flush the connection pool
357 Sometimes applications want to flush all the existing connections kept alive.
358 An API could allow a forced flush or just a forced loop that would properly
359 close all connections that have been closed by the server already.
361 1.24 TCP Fast Open for windows
363 libcurl supports the CURLOPT_TCP_FASTOPEN option since 7.49.0 for Linux and
364 Mac OS. Windows supports TCP Fast Open starting with Windows 10, version 1607
365 and we should add support for it.
367 2. libcurl - multi interface
369 2.1 More non-blocking
371 Make sure we don't ever loop because of non-blocking sockets returning
372 EWOULDBLOCK or similar. Blocking cases include:
374 - Name resolves on non-windows unless c-ares or the threaded resolver is used
375 - SOCKS proxy handshakes
378 - The "DONE" operation (post transfer protocol-specific actions) for the
379 protocols SFTP, SMTP, FTP. Fixing Curl_done() for this is a worthy task.
381 2.2 Better support for same name resolves
383 If a name resolve has been initiated for name NN and a second easy handle
384 wants to resolve that name as well, make it wait for the first resolve to end
385 up in the cache instead of doing a second separate resolve. This is
386 especially needed when adding many simultaneous handles using the same host
387 name when the DNS resolver can get flooded.
389 2.3 Non-blocking curl_multi_remove_handle()
391 The multi interface has a few API calls that assume a blocking behavior, like
392 add_handle() and remove_handle() which limits what we can do internally. The
393 multi API need to be moved even more into a single function that "drives"
394 everything in a non-blocking manner and signals when something is done. A
395 remove or add would then only ask for the action to get started and then
396 multi_perform() etc still be called until the add/remove is completed.
398 2.4 Split connect and authentication process
400 The multi interface treats the authentication process as part of the connect
401 phase. As such any failures during authentication won't trigger the relevant
402 QUIT or LOGOFF for protocols such as IMAP, POP3 and SMTP.
404 2.5 Edge-triggered sockets should work
406 The multi_socket API should work with edge-triggered socket events. One of
407 the internal actions that need to be improved for this to work perfectly is
408 the 'maxloops' handling in transfer.c:readwrite_data().
412 3.2 Provide cmake config-file
414 A config-file package is a set of files provided by us to allow applications
415 to write cmake scripts to find and use libcurl easier. See
416 https://github.com/curl/curl/issues/885
422 HOST is a command for a client to tell which host name to use, to offer FTP
423 servers named-based virtual hosting:
425 https://tools.ietf.org/html/rfc7151
427 4.2 Alter passive/active on failure and retry
429 When trying to connect passively to a server which only supports active
430 connections, libcurl returns CURLE_FTP_WEIRD_PASV_REPLY and closes the
431 connection. There could be a way to fallback to an active connection (and
432 vice versa). https://curl.haxx.se/bug/feature.cgi?id=1754793
434 4.3 Earlier bad letter detection
436 Make the detection of (bad) %0d and %0a codes in FTP URL parts earlier in the
437 process to avoid doing a resolve and connect in vain.
439 4.4 REST for large files
441 REST fix for servers not behaving well on >2GB requests. This should fail if
442 the server doesn't set the pointer to the requested index. The tricky
443 (impossible?) part is to figure out if the server did the right thing or not.
447 FTP ASCII transfers do not follow RFC959. They don't convert the data
450 4.6 GSSAPI via Windows SSPI
452 In addition to currently supporting the SASL GSSAPI mechanism (Kerberos V5)
453 via third-party GSS-API libraries, such as Heimdal or MIT Kerberos, also add
454 support for GSSAPI authentication via Windows SSPI.
456 4.7 STAT for LIST without data connection
458 Some FTP servers allow STAT for listing directories instead of using LIST, and
459 the response is then sent over the control connection instead of as the
460 otherwise usedw data connection: http://www.nsftools.com/tips/RawFTP.htm#STAT
462 This is not detailed in any FTP specification.
466 5.1 Better persistency for HTTP 1.0
468 "Better" support for persistent connections over HTTP 1.0
469 https://curl.haxx.se/bug/feature.cgi?id=1089001
471 5.2 support FF3 sqlite cookie files
473 Firefox 3 is changing from its former format to a a sqlite database instead.
474 We should consider how (lib)curl can/should support this.
475 https://curl.haxx.se/bug/feature.cgi?id=1871388
477 5.3 Rearrange request header order
479 Server implementors often make an effort to detect browser and to reject
480 clients it can detect to not match. One of the last details we cannot yet
481 control in libcurl's HTTP requests, which also can be exploited to detect
482 that libcurl is in fact used even when it tries to impersonate a browser, is
483 the order of the request headers. I propose that we introduce a new option in
484 which you give headers a value, and then when the HTTP request is built it
485 sorts the headers based on that number. We could then have internally created
486 headers use a default value so only headers that need to be moved have to be
489 5.4 HTTP Digest using SHA-256
491 RFC 7616 introduces an update to the HTTP Digest authentication
492 specification, which amongst other thing defines how new digest algorithms
493 can be used instead of MD5 which is considered old and not recommended.
495 See https://tools.ietf.org/html/rfc7616 and
496 https://github.com/curl/curl/issues/1018
500 Add the ability to specify the preferred authentication mechanism to use by
501 using ;auth=<mech> in the login part of the URL.
505 http://test:pass;auth=NTLM@example.com would be equivalent to specifying --user
506 test:pass;auth=NTLM or --user test:pass --ntlm from the command line.
508 Additionally this should be implemented for proxy base URLs as well.
510 5.6 Refuse "downgrade" redirects
512 See https://github.com/curl/curl/issues/226
514 Consider a way to tell curl to refuse to "downgrade" protocol with a redirect
515 and/or possibly a bit that refuses redirect to change protocol completely.
517 5.7 Brotli compression
519 Brotli compression performs better than gzip and is being implemented by
520 browsers and servers widely. The algorithm: https://github.com/google/brotli
521 The Firefox bug: https://bugzilla.mozilla.org/show_bug.cgi?id=366559
525 The standardization process of QUIC has been taken to the IETF and can be
526 followed on the [IETF QUIC Mailing
527 list](https://www.ietf.org/mailman/listinfo/quic). I'd like us to get on the
528 bandwagon. Ideally, this would be done with a separate library/project to
529 handle the binary/framing layer in a similar fashion to how HTTP/2 is
530 implemented. This, to allow other projects to benefit from the work and to
531 thus broaden the interest and chance of others to participate.
533 5.10 Leave secure cookies alone
535 Non-secure origins (HTTP sites) should not be allowed to set or modify
536 cookies with the 'secure' property:
538 https://tools.ietf.org/html/draft-ietf-httpbis-cookie-alone-01
545 Reading input (to send to the remote server) on stdin is a crappy solution for
546 library purposes. We need to invent a good way for the application to be able
547 to provide the data to send.
549 6.2 ditch telnet-specific select
551 Move the telnet support's network select() loop go away and merge the code
552 into the main transfer loop. Until this is done, the multi interface won't
555 6.3 feature negotiation debug data
557 Add telnet feature negotiation data to the debug callback as header data.
564 Add support for pipelining emails.
566 7.2 Enhanced capability support
568 Add the ability, for an application that uses libcurl, to obtain the list of
569 capabilities returned from the EHLO command.
571 7.3 Add CURLOPT_MAIL_CLIENT option
573 Rather than use the URL to specify the mail client string to present in the
574 HELO and EHLO commands, libcurl should support a new CURLOPT specifically for
575 specifying this data as the URL is non-standard and to be honest a bit of a
578 Please see the following thread for more information:
579 https://curl.haxx.se/mail/lib-2012-05/0178.html
586 Add support for pipelining commands.
588 8.2 Enhanced capability support
590 Add the ability, for an application that uses libcurl, to obtain the list of
591 capabilities returned from the CAPA command.
595 9.1 Enhanced capability support
597 Add the ability, for an application that uses libcurl, to obtain the list of
598 capabilities returned from the CAPABILITY command.
602 10.1 SASL based authentication mechanisms
604 Currently the LDAP module only supports ldap_simple_bind_s() in order to bind
605 to an LDAP server. However, this function sends username and password details
606 using the simple authentication mechanism (as clear text). However, it should
607 be possible to use ldap_bind_s() instead specifying the security context
608 information ourselves.
612 11.1 File listing support
614 Add support for listing the contents of a SMB share. The output should probably
615 be the same as/similar to FTP.
617 11.2 Honor file timestamps
619 The timestamp of the transferred file should reflect that of the original file.
623 Currently the SMB authentication uses NTLMv1.
625 11.4 Create remote directories
627 Support for creating remote directories when uploading a file to a directory
628 that doesn't exist on the server, just like --ftp-create-dirs.
634 There's no RFC for the protocol or an URI/URL format. An implementation
635 should most probably use an existing rsync library, such as librsync.
639 13.1 Disable specific versions
641 Provide an option that allows for disabling specific SSL versions, such as
642 SSLv2 https://curl.haxx.se/bug/feature.cgi?id=1767276
644 13.2 Provide mutex locking API
646 Provide a libcurl API for setting mutex callbacks in the underlying SSL
647 library, so that the same application code can use mutex-locking
648 independently of OpenSSL or GnutTLS being used.
650 13.3 Evaluate SSL patches
652 Evaluate/apply Gertjan van Wingerde's SSL patches:
653 https://curl.haxx.se/mail/lib-2004-03/0087.html
655 13.4 Cache/share OpenSSL contexts
657 "Look at SSL cafile - quick traces look to me like these are done on every
658 request as well, when they should only be necessary once per SSL context (or
659 once per handle)". The major improvement we can rather easily do is to make
660 sure we don't create and kill a new SSL "context" for every request, but
661 instead make one for every connection and re-use that SSL context in the same
662 style connections are re-used. It will make us use slightly more memory but
663 it will libcurl do less creations and deletions of SSL contexts.
665 Technically, the "caching" is probably best implemented by getting added to
666 the share interface so that easy handles who want to and can reuse the
667 context specify that by sharing with the right properties set.
669 https://github.com/curl/curl/issues/1110
671 13.5 Export session ids
673 Add an interface to libcurl that enables "session IDs" to get
674 exported/imported. Cris Bailiff said: "OpenSSL has functions which can
675 serialise the current SSL state to a buffer of your choice, and recover/reset
676 the state from such a buffer at a later date - this is used by mod_ssl for
677 apache to implement and SSL session ID cache".
679 13.6 Provide callback for cert verification
681 OpenSSL supports a callback for customised verification of the peer
682 certificate, but this doesn't seem to be exposed in the libcurl APIs. Could
683 it be? There's so much that could be done if it were!
685 13.7 improve configure --with-ssl
687 make the configure --with-ssl option first check for OpenSSL, then GnuTLS,
692 DNS-Based Authentication of Named Entities (DANE) is a way to provide SSL
693 keys and certs over DNS using DNSSEC as an alternative to the CA model.
694 https://www.rfc-editor.org/rfc/rfc6698.txt
696 An initial patch was posted by Suresh Krishnaswamy on March 7th 2013
697 (https://curl.haxx.se/mail/lib-2013-03/0075.html) but it was a too simple
698 approach. See Daniel's comments:
699 https://curl.haxx.se/mail/lib-2013-03/0103.html . libunbound may be the
700 correct library to base this development on.
702 Björn Stenberg wrote a separate initial take on DANE that was never
705 13.10 Support SSLKEYLOGFILE
707 When used, Firefox and Chrome dumps their master TLS keys to the file name
708 this environment variable specifies. This allows tools like for example
709 Wireshark to capture and decipher TLS traffic to/from those clients. libcurl
710 could be made to support this more widely (presumably this already works when
711 built with NSS). Peter Wu made a OpenSSL preload to make possible that can be
712 used as inspiration and guidance
713 https://git.lekensteyn.nl/peter/wireshark-notes/tree/src/sslkeylog.c
715 13.11 Support intermediate & root pinning for PINNEDPUBLICKEY
717 CURLOPT_PINNEDPUBLICKEY does not consider the hashes of intermediate & root
718 certificates when comparing the pinned keys. Therefore it is not compatible
719 with "HTTP Public Key Pinning" as there also intermediate and root certificates
720 can be pinned. This is very useful as it prevents webadmins from "locking
721 themself out of their servers".
723 Adding this feature would make curls pinning 100% compatible to HPKP and allow
724 more flexible pinning.
728 "HTTP Strict Transport Security" is TOFU (trust on first use), time-based
729 features indicated by a HTTP header send by the webserver. It is widely used
730 in browsers and it's purpose is to prevent insecure HTTP connections after
731 a previous HTTPS connection. It protects against SSLStripping attacks.
733 Doc: https://developer.mozilla.org/en-US/docs/Web/Security/HTTP_strict_transport_security
734 RFC 6797: https://tools.ietf.org/html/rfc6797
738 "HTTP Public Key Pinning" is TOFU (trust on first use), time-based
739 features indicated by a HTTP header send by the webserver. It's purpose is
740 to prevent Man-in-the-middle attacks by trusted CAs by allowing webadmins
741 to specify which CAs/certificates/public keys to trust when connection to
744 It can be build based on PINNEDPUBLICKEY.
746 Wikipedia: https://en.wikipedia.org/wiki/HTTP_Public_Key_Pinning
747 OWASP: https://www.owasp.org/index.php/Certificate_and_Public_Key_Pinning
748 Doc: https://developer.mozilla.org/de/docs/Web/Security/Public_Key_Pinning
749 RFC: https://tools.ietf.org/html/draft-ietf-websec-key-pinning-21
753 14.1 SSL engine stuff
755 Is this even possible?
757 14.2 check connection
759 Add a way to check if the connection seems to be alive, to correspond to the
760 SSL_peak() way we use with OpenSSL.
764 15.1 Add support for client certificate authentication
766 WinSSL/SChannel currently makes use of the OS-level system and user
767 certificate and private key stores. This does not allow the application
768 or the user to supply a custom client certificate using curl or libcurl.
770 Therefore support for the existing -E/--cert and --key options should be
771 implemented by supplying a custom certificate to the SChannel APIs, see:
772 - Getting a Certificate for Schannel
773 https://msdn.microsoft.com/en-us/library/windows/desktop/aa375447.aspx
775 15.2 Add support for custom server certificate validation
777 WinSSL/SChannel currently makes use of the OS-level system and user
778 certificate trust store. This does not allow the application or user to
779 customize the server certificate validation process using curl or libcurl.
781 Therefore support for the existing --cacert or --capath options should be
782 implemented by supplying a custom certificate to the SChannel APIs, see:
783 - Getting a Certificate for Schannel
784 https://msdn.microsoft.com/en-us/library/windows/desktop/aa375447.aspx
786 15.3 Add support for the --ciphers option
788 The cipher suites used by WinSSL/SChannel are configured on an OS-level
789 instead of an application-level. This does not allow the application or
790 the user to customize the configured cipher suites using curl or libcurl.
792 Therefore support for the existing --ciphers option should be implemented
793 by mapping the OpenSSL/GnuTLS cipher suites to the SChannel APIs, see
794 - Specifying Schannel Ciphers and Cipher Strengths
795 https://msdn.microsoft.com/en-us/library/windows/desktop/aa380161.aspx
799 16.1 Other authentication mechanisms
801 Add support for other authentication mechanisms such as OLP,
802 GSS-SPNEGO and others.
804 16.2 Add QOP support to GSSAPI authentication
806 Currently the GSSAPI authentication only supports the default QOP of auth
807 (Authentication), whilst Kerberos V5 supports both auth-int (Authentication
808 with integrity protection) and auth-conf (Authentication with integrity and
811 16.3 Support binary messages (i.e.: non-base64)
813 Mandatory to support LDAP SASL authentication.
820 SSH is a perfectly fine multiplexed protocols which would allow libcurl to do
821 multiple parallel transfers from the same host using the same connection,
822 much in the same spirit as HTTP/2 does. libcurl however does not take
823 advantage of that ability but will instead always create a new connection for
824 new transfers even if an existing connection already exists to the host.
826 To fix this, libcurl would have to detect an existing connection and "attach"
827 the new transfer to the existing one.
829 17.2 SFTP performance
831 libcurl's SFTP transfer performance is sub par and can be improved, mostly by
832 the approach mentioned in "1.6 Modified buffer size approach".
834 17.3 Support better than MD5 hostkey hash
836 libcurl offers the CURLOPT_SSH_HOST_PUBLIC_KEY_MD5 option for verifying the
837 server's key. MD5 is generally being deprecated so we should implement
838 support for stronger hashing algorithms. libssh2 itself is what provides this
839 underlying functionality and it supports at least SHA-1 as an alternative.
840 SHA-1 is also being deprecated these days so we should consider workign with
841 libssh2 to instead offer support for SHA-256 or similar.
843 17.4 Support CURLOPT_PREQUOTE
845 The two other QUOTE options are supported for SFTP, but this was left out for
848 18. Command line tool
852 "curl --sync http://example.com/feed[1-100].rss" or
853 "curl --sync http://example.net/{index,calendar,history}.html"
855 Downloads a range or set of URLs using the remote name, but only if the
856 remote file is newer than the local file. A Last-Modified HTTP date header
857 should also be used to set the mod date on the downloaded file.
861 Globbing support for -d and -F, as in 'curl -d "name=foo[0-9]" URL'.
862 This is easily scripted though.
864 18.3 prevent file overwriting
866 Add an option that prevents curl from overwriting existing local files. When
867 used, and there already is an existing file with the target file name
868 (either -O or -o), a number should be appended (and increased if already
869 existing). So that index.html becomes first index.html.1 and then
872 18.4 simultaneous parallel transfers
874 The client could be told to use maximum N simultaneous parallel transfers and
875 then just make sure that happens. It should of course not make more than one
876 connection to the same remote host. This would require the client to use the
877 multi interface. https://curl.haxx.se/bug/feature.cgi?id=1558595
879 Using the multi interface would also allow properly using parallel transfers
880 with HTTP/2 and supporting HTTP/2 server push from the command line.
882 18.6 warning when setting an option
884 Display a warning when libcurl returns an error when setting an option.
885 This can be useful to tell when support for a particular feature hasn't been
886 compiled into the library.
888 18.8 offer color-coded HTTP header output
890 By offering different color output on the header name and the header
891 contents, they could be made more readable and thus help users working on
894 18.9 Choose the name of file in braces for complex URLs
896 When using braces to download a list of URLs and you use complicated names
897 in the list of alternatives, it could be handy to allow curl to use other
900 Consider a way to offer that. Possibly like
901 {partURL1:name1,partURL2:name2,partURL3:name3} where the name following the
902 colon is the output name.
904 See https://github.com/curl/curl/issues/221
906 18.10 improve how curl works in a windows console window
908 If you pull the scrollbar when transferring with curl in a Windows console
909 window, the transfer is interrupted and can get disconnected. This can
910 probably be improved. See https://github.com/curl/curl/issues/322
912 18.11 -w output to stderr
914 -w is quite useful, but not to those of us who use curl without -o or -O
915 (such as for scripting through a higher level language). It would be nice to
916 have an option that is exactly like -w but sends it to stderr
917 instead. Proposed name: --write-stderr. See
918 https://github.com/curl/curl/issues/613
920 18.12 keep running, read instructions from pipe/socket
922 Provide an option that makes curl not exit after the last URL (or even work
923 without a given URL), and then make it read instructions passed on a pipe or
924 over a socket to make further instructions so that a second subsequent curl
925 invoke can talk to the still running instance and ask for transfers to get
926 done, and thus maintain its connection pool, DNS cache and more.
928 18.13 support metalink in http headers
930 Curl has support for downloading a metalink xml file, processing it, and then
931 downloading the target of the metalink. This is done via the --metalink option.
932 It would be nice if metalink also supported downloading via metalink
933 information that is stored in HTTP headers (RFC 6249). Theoretically this could
934 also be supported with the --metalink option.
936 See https://tools.ietf.org/html/rfc6249
938 See also https://lists.gnu.org/archive/html/bug-wget/2015-06/msg00034.html for
939 an implematation of this in wget.
941 18.14 --fail without --location should treat 3xx as a failure
943 To allow a command line like this to detect a redirect and consider it a
946 curl -v --fail -O https://example.com/curl-7.48.0.tar.gz
948 ... --fail must treat 3xx responses as failures too. The least problematic
949 way to implement this is probably to add that new logic in the command line
950 tool only and not in the underlying CURLOPT_FAILONERROR logic.
952 18.15 --retry should resume
954 When --retry is used and curl actually retries transfer, it should use the
955 already transferred data and do a resumed transfer for the rest (when
956 possible) so that it doesn't have to transfer the same data again that was
957 already transferred before the retry.
959 See https://github.com/curl/curl/issues/1084
961 18.16 send only part of --data
963 When the user only wants to send a small piece of the data provided with
964 --data or --data-binary, like when that data is a huge file, consider a way
965 to specify that curl should only send a piece of that. One suggested syntax
966 would be: "--data-binary @largefile.zip!1073741823-2147483647".
968 See https://github.com/curl/curl/issues/1200
970 18.17 consider file name from the redirected URL with -O ?
972 When a user gives a URL and uses -O, and curl follows a redirect to a new
973 URL, the file name is not extracted and used from the newly redirected-to URL
974 even if the new URL may have a much more sensible file name.
976 This is clearly documented and helps for security since there's no surprise
977 to users which file name that might get overwritten. But maybe a new option
978 could allow for this or maybe -J should imply such a treatment as well as -J
979 already allows for the server to decide what file name to use so it already
980 provides the "may overwrite any file" risk.
982 This is extra tricky if the original URL has no file name part at all since
983 then the current code path will error out with an error message, and we can't
984 *know* already at that point if curl will be redirected to a URL that has a
987 See https://github.com/curl/curl/issues/1241
993 Consider extending 'roffit' to produce decent ASCII output, and use that
994 instead of (g)nroff when building src/tool_hugehelp.c
996 19.2 Enable PIE and RELRO by default
998 Especially when having programs that execute curl via the command line, PIE
999 renders the exploitation of memory corruption vulnerabilities a lot more
1000 difficult. This can be attributed to the additional information leaks being
1001 required to conduct a successful attack. RELRO, on the other hand, masks
1002 different binary sections like the GOT as read-only and thus kills a handful
1003 of techniques that come in handy when attackers are able to arbitrarily
1004 overwrite memory. A few tests showed that enabling these features had close
1005 to no impact, neither on the performance nor on the general functionality of
1013 Make our own version of stunnel for simple port forwarding to enable HTTPS
1014 and FTP-SSL tests without the stunnel dependency, and it could allow us to
1015 provide test tools built with either OpenSSL or GnuTLS
1017 20.2 nicer lacking perl message
1019 If perl wasn't found by the configure script, don't attempt to run the tests
1020 but explain something nice why it doesn't.
1022 20.3 more protocols supported
1024 Extend the test suite to include more protocols. The telnet could just do FTP
1025 or http operations (for which we have test servers).
1027 20.4 more platforms supported
1029 Make the test suite work on more platforms. OpenBSD and Mac OS. Remove
1030 fork()s and it should become even more portable.
1032 20.5 Add support for concurrent connections
1034 Tests 836, 882 and 938 were designed to verify that separate connections aren't
1035 used when using different login credentials in protocols that shouldn't re-use
1036 a connection under such circumstances.
1038 Unfortunately, ftpserver.pl doesn't appear to support multiple concurrent
1039 connections. The read while() loop seems to loop until it receives a disconnect
1040 from the client, where it then enters the waiting for connections loop. When
1041 the client opens a second connection to the server, the first connection hasn't
1042 been dropped (unless it has been forced - which we shouldn't do in these tests)
1043 and thus the wait for connections loop is never entered to receive the second
1046 20.6 Use the RFC6265 test suite
1048 A test suite made for HTTP cookies (RFC 6265) by Adam Barth is available at
1049 https://github.com/abarth/http-state/tree/master/tests
1051 It'd be really awesome if someone would write a script/setup that would run
1052 curl with that test suite and detect deviances. Ideally, that would even be
1053 incorporated into our regular test suite.
1056 21. Next SONAME bump
1058 21.1 http-style HEAD output for FTP
1060 #undef CURL_FTP_HTTPSTYLE_HEAD in lib/ftp.c to remove the HTTP-style headers
1061 from being output in NOBODY requests over FTP
1063 21.2 combine error codes
1065 Combine some of the error codes to remove duplicates. The original
1066 numbering should not be changed, and the old identifiers would be
1067 macroed to the new ones in an CURL_NO_OLDIES section to help with
1068 backward compatibility.
1070 Candidates for removal and their replacements:
1072 CURLE_FILE_COULDNT_READ_FILE => CURLE_REMOTE_FILE_NOT_FOUND
1074 CURLE_FTP_COULDNT_RETR_FILE => CURLE_REMOTE_FILE_NOT_FOUND
1076 CURLE_FTP_COULDNT_USE_REST => CURLE_RANGE_ERROR
1078 CURLE_FUNCTION_NOT_FOUND => CURLE_FAILED_INIT
1080 CURLE_LDAP_INVALID_URL => CURLE_URL_MALFORMAT
1082 CURLE_TFTP_NOSUCHUSER => CURLE_TFTP_ILLEGAL
1084 CURLE_TFTP_NOTFOUND => CURLE_REMOTE_FILE_NOT_FOUND
1086 CURLE_TFTP_PERM => CURLE_REMOTE_ACCESS_DENIED
1088 21.3 extend CURLOPT_SOCKOPTFUNCTION prototype
1090 The current prototype only provides 'purpose' that tells what the
1091 connection/socket is for, but not any protocol or similar. It makes it hard
1092 for applications to differentiate on TCP vs UDP and even HTTP vs FTP and
1095 22. Next major release
1097 22.1 cleanup return codes
1099 curl_easy_cleanup() returns void, but curl_multi_cleanup() returns a
1100 CURLMcode. These should be changed to be the same.
1102 22.2 remove obsolete defines
1104 remove obsolete defines from curl/curl.h
1108 make several functions use size_t instead of int in their APIs
1110 22.4 remove several functions
1112 remove the following functions from the public API:
1116 curl_mprintf (and variations)
1122 They will instead become curlx_ - alternatives. That makes the curl app
1123 still capable of using them, by building with them from source.
1125 These functions have no purpose anymore:
1129 curl_multi_socket_all
1131 22.5 remove CURLOPT_FAILONERROR
1133 Remove support for CURLOPT_FAILONERROR, it has gotten too kludgy and weird
1134 internally. Let the app judge success or not for itself.
1136 22.6 remove CURLOPT_DNS_USE_GLOBAL_CACHE
1138 Remove support for a global DNS cache. Anything global is silly, and we
1139 already offer the share interface for the same functionality but done
1142 22.7 remove progress meter from libcurl
1144 The internally provided progress meter output doesn't belong in the library.
1145 Basically no application wants it (apart from curl) but instead applications
1146 can and should do their own progress meters using the progress callback.
1148 The progress callback should then be bumped as well to get proper 64bit
1149 variable types passed to it instead of doubles so that big files work
1152 22.8 remove 'curl_httppost' from public
1154 curl_formadd() was made to fill in a public struct, but the fact that the
1155 struct is public is never really used by application for their own advantage
1156 but instead often restricts how the form functions can or can't be modified.
1158 Changing them to return a private handle will benefit the implementation and
1159 allow us much greater freedoms while still maintaining a solid API and ABI.