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 Allow SSL (HTTPS) to proxy
27 1.9 Cache negative name resolves
29 1.11 minimize dependencies with dynamicly loaded modules
30 1.12 have form functions use CURL handle argument
31 1.14 Typesafe curl_easy_setopt()
32 1.15 Monitor connections in the connection pool
33 1.16 Try to URL encode given URL
34 1.17 Add support for IRIs
35 1.18 try next proxy if one doesn't work
36 1.19 Timeout idle connections from the pool
37 1.20 SRV and URI DNS records
38 1.21 API for URL parsing/splitting
39 1.23 Offer API to flush the connection pool
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 3.1 Update date and version in man pages
50 3.2 Provide cmake config-file
54 4.2 Alter passive/active on failure and retry
55 4.3 Earlier bad letter detection
56 4.4 REST for large files
58 4.6 GSSAPI via Windows SSPI
59 4.7 STAT for LIST without data connection
62 5.1 Better persistency for HTTP 1.0
63 5.2 support FF3 sqlite cookie files
64 5.3 Rearrange request header order
65 5.4 Use huge HTTP/2 windows
67 5.6 Refuse "downgrade" redirects
68 5.7 Brotli compression
73 6.2 ditch telnet-specific select
74 6.3 feature negotiation debug data
75 6.4 send data in chunks
79 7.2 Enhanced capability support
80 7.3 Add CURLOPT_MAIL_CLIENT option
84 8.2 Enhanced capability support
87 9.1 Enhanced capability support
90 10.1 SASL based authentication mechanisms
93 11.1 File listing support
94 11.2 Honor file timestamps
96 11.4 Create remote directories
102 13.1 Disable specific versions
103 13.2 Provide mutex locking API
104 13.3 Evaluate SSL patches
105 13.4 Cache OpenSSL contexts
106 13.5 Export session ids
107 13.6 Provide callback for cert verification
108 13.7 improve configure --with-ssl
110 13.9 Support TLS v1.3
113 14.1 SSL engine stuff
114 14.2 check connection
117 15.1 Add support for client certificate authentication
118 15.2 Add support for custom server certificate validation
119 15.3 Add support for the --ciphers option
122 16.1 Other authentication mechanisms
123 16.2 Add QOP support to GSSAPI authentication
124 16.3 Support binary messages (i.e.: non-base64)
128 17.2 SFTP performance
130 18. Command line tool
133 18.3 prevent file overwriting
134 18.4 simultaneous parallel transfers
135 18.5 provide formpost headers
136 18.6 warning when setting an option
137 18.7 warning when sending binary output to terminal
138 18.8 offer color-coded HTTP header output
139 18.9 Choose the name of file in braces for complex URLs
140 18.10 improve how curl works in a windows console window
141 18.11 -w output to stderr
142 18.12 keep running, read instructions from pipe/socket
143 18.13 support metalink in http headers
144 18.14 --fail without --location should treat 3xx as a failure
151 20.2 nicer lacking perl message
152 20.3 more protocols supported
153 20.4 more platforms supported
154 20.5 Add support for concurrent connections
155 20.6 Use the RFC6265 test suite
158 21.1 http-style HEAD output for FTP
159 21.2 combine error codes
160 21.3 extend CURLOPT_SOCKOPTFUNCTION prototype
162 22. Next major release
163 22.1 cleanup return codes
164 22.2 remove obsolete defines
166 22.4 remove several functions
167 22.5 remove CURLOPT_FAILONERROR
168 22.6 remove CURLOPT_DNS_USE_GLOBAL_CACHE
169 22.7 remove progress meter from libcurl
170 22.8 remove 'curl_httppost' from public
172 ==============================================================================
176 1.2 More data sharing
178 curl_share_* functions already exist and work, and they can be extended to
179 share more. For example, enable sharing of the ares channel and the
184 Use 'struct lifreq' and SIOCGLIFADDR instead of 'struct ifreq' and
185 SIOCGIFADDR on newer Solaris versions as they claim the latter is obsolete.
186 To support IPv6 interface addresses for network interfaces properly.
188 1.4 signal-based resolver timeouts
190 libcurl built without an asynchronous resolver library uses alarm() to time
191 out DNS lookups. When a timeout occurs, this causes libcurl to jump from the
192 signal handler back into the library with a sigsetjmp, which effectively
193 causes libcurl to continue running within the signal handler. This is
194 non-portable and could cause problems on some platforms. A discussion on the
195 problem is available at https://curl.haxx.se/mail/lib-2008-09/0197.html
197 Also, alarm() provides timeout resolution only to the nearest second. alarm
198 ought to be replaced by setitimer on systems that support it.
200 1.5 get rid of PATH_MAX
202 Having code use and rely on PATH_MAX is not nice:
203 http://insanecoding.blogspot.com/2007/11/pathmax-simply-isnt.html
205 Currently the SSH based code uses it a bit, but to remove PATH_MAX from there
206 we need libssh2 to properly tell us when we pass in a too small buffer and
207 its current API (as of libssh2 1.2.7) doesn't.
209 1.6 Modified buffer size approach
211 Current libcurl allocates a fixed 16K size buffer for download and an
212 additional 16K for upload. They are always unconditionally part of the easy
213 handle. If CRLF translations are requested, an additional 32K "scratch
214 buffer" is allocated. A total of 64K transfer buffers in the worst case.
216 First, while the handles are not actually in use these buffers could be freed
217 so that lingering handles just kept in queues or whatever waste less memory.
219 Secondly, SFTP is a protocol that needs to handle many ~30K blocks at once
220 since each need to be individually acked and therefore libssh2 must be
221 allowed to send (or receive) many separate ones in parallel to achieve high
222 transfer speeds. A current libcurl build with a 16K buffer makes that
223 impossible, but one with a 512K buffer will reach MUCH faster transfers. But
224 allocating 512K unconditionally for all buffers just in case they would like
225 to do fast SFTP transfers at some point is not a good solution either.
227 Dynamically allocate buffer size depending on protocol in use in combination
228 with freeing it after each individual transfer? Other suggestions?
230 1.7 Detect when called from within callbacks
232 We should set a state variable before calling callbacks, so that we
233 subsequently can add code within libcurl that returns error if called within
234 callbacks for when that's not supported.
236 1.8 Allow SSL (HTTPS) to proxy
238 To prevent local users from snooping on your traffic to the proxy. Supported
239 by Firefox and Chrome already:
240 https://www.chromium.org/developers/design-documents/secure-web-proxy
242 See this stale work in progress branch:
243 https://github.com/curl/curl/tree/HTTPS-proxy based on this PR:
244 https://github.com/curl/curl/pull/305
246 1.9 Cache negative name resolves
248 A name resolve that has failed is likely to fail when made again within a
249 short period of time. Currently we only cache positive responses.
251 1.10 Support IDNA2008
253 International Domain Names are supported in libcurl since years back, powered
254 by libidn. libidn implements IDNA2003 which has been superseded by IDNA2008.
255 libidn2 is an existing library offering support for IDNA2008.
257 1.11 minimize dependencies with dynamicly loaded modules
259 We can create a system with loadable modules/plug-ins, where these modules
260 would be the ones that link to 3rd party libs. That would allow us to avoid
261 having to load ALL dependencies since only the necessary ones for this
262 app/invoke/used protocols would be necessary to load. See
263 https://github.com/curl/curl/issues/349
265 1.12 have form functions use CURL handle argument
267 curl_formadd() and curl_formget() both currently have no CURL handle
268 argument, but both can use a callback that is set in the easy handle, and
269 thus curl_formget() with callback cannot function without first having
270 curl_easy_perform() (or similar) called - which is hard to grasp and a design
273 The curl_formadd() design can probably also be reconsidered to make it easier
274 to use and less error-prone. Probably easiest by splitting it into several
277 1.14 Typesafe curl_easy_setopt()
279 One of the most common problems in libcurl using applications is the lack of
280 type checks for curl_easy_setopt() which happens because it accepts varargs
281 and thus can take any type.
283 One possible solution to this is to introduce a few different versions of the
284 setopt version for the different kinds of data you can set.
286 curl_easy_set_num() - sets a long value
288 curl_easy_set_large() - sets a curl_off_t value
290 curl_easy_set_ptr() - sets a pointer
292 curl_easy_set_cb() - sets a callback PLUS its callback data
294 1.15 Monitor connections in the connection pool
296 libcurl's connection cache or pool holds a number of open connections for the
297 purpose of possible subsequent connection reuse. It may contain a few up to a
298 significant amount of connections. Currently, libcurl leaves all connections
299 as they are and first when a connection is iterated over for matching or
300 reuse purpose it is verified that it is still alive.
302 Those connections may get closed by the server side for idleness or they may
303 get a HTTP/2 ping from the peer to verify that they're still alive. By adding
304 monitoring of the connections while in the pool, libcurl can detect dead
305 connections (and close them) better and earlier, and it can handle HTTP/2
306 pings to keep such ones alive even when not actively doing transfers on them.
308 1.16 Try to URL encode given URL
310 Given a URL that for example contains spaces, libcurl could have an option
311 that would try somewhat harder than it does now and convert spaces to %20 and
312 perhaps URL encoded byte values over 128 etc (basically do what the redirect
313 following code already does).
315 https://github.com/curl/curl/issues/514
317 1.17 Add support for IRIs
319 IRIs (RFC 3987) allow localized, non-ascii, names in the URL. To properly
320 support this, curl/libcurl would need to translate/encode the given input
321 from the input string encoding into percent encoded output "over the wire".
323 To make that work smoothly for curl users even on Windows, curl would
324 probably need to be able to convert from several input encodings.
326 1.18 try next proxy if one doesn't work
328 Allow an application to specify a list of proxies to try, and failing to
329 connect to the first go on and try the next instead until the list is
330 exhausted. Browsers support this feature at least when they specify proxies
333 https://github.com/curl/curl/issues/896
335 1.19 Timeout idle connections from the pool
337 libcurl currently keeps connections in its connection pool for an indefinite
338 period of time, until it either gets reused, gets noticed that it has been
339 closed by the server or gets pruned to make room for a new connection.
341 To reduce overhead (especially for when we add monitoring of the connections
342 in the pool), we should introduce a timeout so that connections that have
343 been idle for N seconds get closed.
345 1.20 SRV and URI DNS records
347 Offer support for resolving SRV and URI DNS records for libcurl to know which
348 server to connect to for various protocols (including HTTP!).
350 1.21 API for URL parsing/splitting
352 libcurl has always parsed URLs internally and never exposed any API or
353 features to allow applications to do it. Still most or many applications
354 using libcurl need that ability. In polls to users, we've learned that many
355 libcurl users would like to see and use such an API.
357 1.23 Offer API to flush the connection pool
359 Sometimes applications want to flush all the existing connections kept alive.
360 An API could allow a forced flush or just a forced loop that would properly
361 close all connections that have been closed by the server already.
364 2. libcurl - multi interface
366 2.1 More non-blocking
368 Make sure we don't ever loop because of non-blocking sockets returning
369 EWOULDBLOCK or similar. Blocking cases include:
371 - Name resolves on non-windows unless c-ares is used
372 - NSS SSL connections
373 - HTTP proxy CONNECT operations
374 - SOCKS proxy handshakes
377 - The "DONE" operation (post transfer protocol-specific actions) for the
378 protocols SFTP, SMTP, FTP. Fixing Curl_done() for this is a worthy task.
380 2.2 Better support for same name resolves
382 If a name resolve has been initiated for name NN and a second easy handle
383 wants to resolve that name as well, make it wait for the first resolve to end
384 up in the cache instead of doing a second separate resolve. This is
385 especially needed when adding many simultaneous handles using the same host
386 name when the DNS resolver can get flooded.
388 2.3 Non-blocking curl_multi_remove_handle()
390 The multi interface has a few API calls that assume a blocking behavior, like
391 add_handle() and remove_handle() which limits what we can do internally. The
392 multi API need to be moved even more into a single function that "drives"
393 everything in a non-blocking manner and signals when something is done. A
394 remove or add would then only ask for the action to get started and then
395 multi_perform() etc still be called until the add/remove is completed.
397 2.4 Split connect and authentication process
399 The multi interface treats the authentication process as part of the connect
400 phase. As such any failures during authentication won't trigger the relevant
401 QUIT or LOGOFF for protocols such as IMAP, POP3 and SMTP.
403 2.5 Edge-triggered sockets should work
405 The multi_socket API should work with edge-triggered socket events. One of
406 the internal actions that need to be improved for this to work perfectly is
407 the 'maxloops' handling in transfer.c:readwrite_data().
411 3.1 Update date and version in man pages
413 'maketgz' or another suitable script could update the .TH sections of the man
414 pages at release time to use the current date and curl/libcurl version
417 3.2 Provide cmake config-file
419 A config-file package is a set of files provided by us to allow applications
420 to write cmake scripts to find and use libcurl easier. See
421 https://github.com/curl/curl/issues/885
427 HOST is a command for a client to tell which host name to use, to offer FTP
428 servers named-based virtual hosting:
430 https://tools.ietf.org/html/rfc7151
432 4.2 Alter passive/active on failure and retry
434 When trying to connect passively to a server which only supports active
435 connections, libcurl returns CURLE_FTP_WEIRD_PASV_REPLY and closes the
436 connection. There could be a way to fallback to an active connection (and
437 vice versa). https://curl.haxx.se/bug/feature.cgi?id=1754793
439 4.3 Earlier bad letter detection
441 Make the detection of (bad) %0d and %0a codes in FTP URL parts earlier in the
442 process to avoid doing a resolve and connect in vain.
444 4.4 REST for large files
446 REST fix for servers not behaving well on >2GB requests. This should fail if
447 the server doesn't set the pointer to the requested index. The tricky
448 (impossible?) part is to figure out if the server did the right thing or not.
452 FTP ASCII transfers do not follow RFC959. They don't convert the data
455 4.6 GSSAPI via Windows SSPI
457 In addition to currently supporting the SASL GSSAPI mechanism (Kerberos V5)
458 via third-party GSS-API libraries, such as Heimdal or MIT Kerberos, also add
459 support for GSSAPI authentication via Windows SSPI.
461 4.7 STAT for LIST without data connection
463 Some FTP servers allow STAT for listing directories instead of using LIST, and
464 the response is then sent over the control connection instead of as the
465 otherwise usedw data connection: http://www.nsftools.com/tips/RawFTP.htm#STAT
467 This is not detailed in any FTP specification.
471 5.1 Better persistency for HTTP 1.0
473 "Better" support for persistent connections over HTTP 1.0
474 https://curl.haxx.se/bug/feature.cgi?id=1089001
476 5.2 support FF3 sqlite cookie files
478 Firefox 3 is changing from its former format to a a sqlite database instead.
479 We should consider how (lib)curl can/should support this.
480 https://curl.haxx.se/bug/feature.cgi?id=1871388
482 5.3 Rearrange request header order
484 Server implementors often make an effort to detect browser and to reject
485 clients it can detect to not match. One of the last details we cannot yet
486 control in libcurl's HTTP requests, which also can be exploited to detect
487 that libcurl is in fact used even when it tries to impersonate a browser, is
488 the order of the request headers. I propose that we introduce a new option in
489 which you give headers a value, and then when the HTTP request is built it
490 sorts the headers based on that number. We could then have internally created
491 headers use a default value so only headers that need to be moved have to be
494 5.4 Use huge HTTP/2 windows
496 We're currently using nghttp2's default window size which is terribly small
497 (64K). This becomes a bottle neck over high bandwidth networks. We should
498 instead make the window size to be very big (512MB?) as we really don't do
499 much flow control anyway.
503 Add the ability to specify the preferred authentication mechanism to use by
504 using ;auth=<mech> in the login part of the URL.
508 http://test:pass;auth=NTLM@example.com would be equivalent to specifying --user
509 test:pass;auth=NTLM or --user test:pass --ntlm from the command line.
511 Additionally this should be implemented for proxy base URLs as well.
513 5.6 Refuse "downgrade" redirects
515 See https://github.com/curl/curl/issues/226
517 Consider a way to tell curl to refuse to "downgrade" protocol with a redirect
518 and/or possibly a bit that refuses redirect to change protocol completely.
520 5.7 Brotli compression
522 Compression algorithms that perform better than gzip are being considered for
523 use and inclusion in existing browsers. For example 'brotli'. If servers
524 follow along it is a good reason for us to also allow users to take advantage
525 of this. The algorithm: https://github.com/google/brotli The Firefox bug:
526 https://bugzilla.mozilla.org/show_bug.cgi?id=366559
530 The standardization process of QUIC has been taken to the IETF and can be
531 followed on the [IETF QUIC Mailing
532 list](https://www.ietf.org/mailman/listinfo/quic). I'd like us to get on the
533 bandwagon. Ideally, this would be done with a separate library/project to
534 handle the binary/framing layer in a similar fashion to how HTTP/2 is
535 implemented. This, to allow other projects to benefit from the work and to
536 thus broaden the interest and chance of others to participate.
543 Reading input (to send to the remote server) on stdin is a crappy solution for
544 library purposes. We need to invent a good way for the application to be able
545 to provide the data to send.
547 6.2 ditch telnet-specific select
549 Move the telnet support's network select() loop go away and merge the code
550 into the main transfer loop. Until this is done, the multi interface won't
553 6.3 feature negotiation debug data
555 Add telnet feature negotiation data to the debug callback as header data.
557 6.4 send data in chunks
559 Currently, telnet sends data one byte at a time. This is fine for interactive
560 use, but inefficient for any other. Sent data should be sent in larger
567 Add support for pipelining emails.
569 7.2 Enhanced capability support
571 Add the ability, for an application that uses libcurl, to obtain the list of
572 capabilities returned from the EHLO command.
574 7.3 Add CURLOPT_MAIL_CLIENT option
576 Rather than use the URL to specify the mail client string to present in the
577 HELO and EHLO commands, libcurl should support a new CURLOPT specifically for
578 specifying this data as the URL is non-standard and to be honest a bit of a
581 Please see the following thread for more information:
582 https://curl.haxx.se/mail/lib-2012-05/0178.html
589 Add support for pipelining commands.
591 8.2 Enhanced capability support
593 Add the ability, for an application that uses libcurl, to obtain the list of
594 capabilities returned from the CAPA command.
598 9.1 Enhanced capability support
600 Add the ability, for an application that uses libcurl, to obtain the list of
601 capabilities returned from the CAPABILITY command.
605 10.1 SASL based authentication mechanisms
607 Currently the LDAP module only supports ldap_simple_bind_s() in order to bind
608 to an LDAP server. However, this function sends username and password details
609 using the simple authentication mechanism (as clear text). However, it should
610 be possible to use ldap_bind_s() instead specifying the security context
611 information ourselves.
615 11.1 File listing support
617 Add support for listing the contents of a SMB share. The output should probably
618 be the same as/similar to FTP.
620 11.2 Honor file timestamps
622 The timestamp of the transferred file should reflect that of the original file.
626 Currently the SMB authentication uses NTLMv1.
628 11.4 Create remote directories
630 Support for creating remote directories when uploading a file to a directory
631 that doesn't exist on the server, just like --ftp-create-dirs.
637 There's no RFC for the protocol or an URI/URL format. An implementation
638 should most probably use an existing rsync library, such as librsync.
642 13.1 Disable specific versions
644 Provide an option that allows for disabling specific SSL versions, such as
645 SSLv2 https://curl.haxx.se/bug/feature.cgi?id=1767276
647 13.2 Provide mutex locking API
649 Provide a libcurl API for setting mutex callbacks in the underlying SSL
650 library, so that the same application code can use mutex-locking
651 independently of OpenSSL or GnutTLS being used.
653 13.3 Evaluate SSL patches
655 Evaluate/apply Gertjan van Wingerde's SSL patches:
656 https://curl.haxx.se/mail/lib-2004-03/0087.html
658 13.4 Cache OpenSSL contexts
660 "Look at SSL cafile - quick traces look to me like these are done on every
661 request as well, when they should only be necessary once per SSL context (or
662 once per handle)". The major improvement we can rather easily do is to make
663 sure we don't create and kill a new SSL "context" for every request, but
664 instead make one for every connection and re-use that SSL context in the same
665 style connections are re-used. It will make us use slightly more memory but
666 it will libcurl do less creations and deletions of SSL contexts.
668 13.5 Export session ids
670 Add an interface to libcurl that enables "session IDs" to get
671 exported/imported. Cris Bailiff said: "OpenSSL has functions which can
672 serialise the current SSL state to a buffer of your choice, and recover/reset
673 the state from such a buffer at a later date - this is used by mod_ssl for
674 apache to implement and SSL session ID cache".
676 13.6 Provide callback for cert verification
678 OpenSSL supports a callback for customised verification of the peer
679 certificate, but this doesn't seem to be exposed in the libcurl APIs. Could
680 it be? There's so much that could be done if it were!
682 13.7 improve configure --with-ssl
684 make the configure --with-ssl option first check for OpenSSL, then GnuTLS,
689 DNS-Based Authentication of Named Entities (DANE) is a way to provide SSL
690 keys and certs over DNS using DNSSEC as an alternative to the CA model.
691 https://www.rfc-editor.org/rfc/rfc6698.txt
693 An initial patch was posted by Suresh Krishnaswamy on March 7th 2013
694 (https://curl.haxx.se/mail/lib-2013-03/0075.html) but it was a too simple
695 approach. See Daniel's comments:
696 https://curl.haxx.se/mail/lib-2013-03/0103.html . libunbound may be the
697 correct library to base this development on.
699 Björn Stenberg wrote a separate initial take on DANE that was never
702 13.9 Support TLS v1.3
704 TLS version 1.3 is about to ship and is getting implemented by TLS libraries
705 as we speak. We should start to support the symbol and make sure all backends
706 handle it accordingly, then gradually add support as the TLS libraries add
707 the corresponding support. There may be a need to add some additional options
708 to allow libcurl to take advantage of the new features in 1.3.
713 14.1 SSL engine stuff
715 Is this even possible?
717 14.2 check connection
719 Add a way to check if the connection seems to be alive, to correspond to the
720 SSL_peak() way we use with OpenSSL.
724 15.1 Add support for client certificate authentication
726 WinSSL/SChannel currently makes use of the OS-level system and user
727 certificate and private key stores. This does not allow the application
728 or the user to supply a custom client certificate using curl or libcurl.
730 Therefore support for the existing -E/--cert and --key options should be
731 implemented by supplying a custom certificate to the SChannel APIs, see:
732 - Getting a Certificate for Schannel
733 https://msdn.microsoft.com/en-us/library/windows/desktop/aa375447.aspx
735 15.2 Add support for custom server certificate validation
737 WinSSL/SChannel currently makes use of the OS-level system and user
738 certificate trust store. This does not allow the application or user to
739 customize the server certificate validation process using curl or libcurl.
741 Therefore support for the existing --cacert or --capath options should be
742 implemented by supplying a custom certificate to the SChannel APIs, see:
743 - Getting a Certificate for Schannel
744 https://msdn.microsoft.com/en-us/library/windows/desktop/aa375447.aspx
746 15.3 Add support for the --ciphers option
748 The cipher suites used by WinSSL/SChannel are configured on an OS-level
749 instead of an application-level. This does not allow the application or
750 the user to customize the configured cipher suites using curl or libcurl.
752 Therefore support for the existing --ciphers option should be implemented
753 by mapping the OpenSSL/GnuTLS cipher suites to the SChannel APIs, see
754 - Specifying Schannel Ciphers and Cipher Strengths
755 https://msdn.microsoft.com/en-us/library/windows/desktop/aa380161.aspx
759 16.1 Other authentication mechanisms
761 Add support for other authentication mechanisms such as OLP,
762 GSS-SPNEGO and others.
764 16.2 Add QOP support to GSSAPI authentication
766 Currently the GSSAPI authentication only supports the default QOP of auth
767 (Authentication), whilst Kerberos V5 supports both auth-int (Authentication
768 with integrity protection) and auth-conf (Authentication with integrity and
771 16.3 Support binary messages (i.e.: non-base64)
773 Mandatory to support LDAP SASL authentication.
780 SSH is a perfectly fine multiplexed protocols which would allow libcurl to do
781 multiple parallel transfers from the same host using the same connection,
782 much in the same spirit as HTTP/2 does. libcurl however does not take
783 advantage of that ability but will instead always create a new connection for
784 new transfers even if an existing connection already exists to the host.
786 To fix this, libcurl would have to detect an existing connection and "attach"
787 the new transfer to the existing one.
789 17.2 SFTP performance
791 libcurl's SFTP transfer performance is sub par and can be improved, mostly by
792 the approach mentioned in "1.6 Modified buffer size approach".
794 18. Command line tool
798 "curl --sync http://example.com/feed[1-100].rss" or
799 "curl --sync http://example.net/{index,calendar,history}.html"
801 Downloads a range or set of URLs using the remote name, but only if the
802 remote file is newer than the local file. A Last-Modified HTTP date header
803 should also be used to set the mod date on the downloaded file.
807 Globbing support for -d and -F, as in 'curl -d "name=foo[0-9]" URL'.
808 This is easily scripted though.
810 18.3 prevent file overwriting
812 Add an option that prevents cURL from overwriting existing local files. When
813 used, and there already is an existing file with the target file name
814 (either -O or -o), a number should be appended (and increased if already
815 existing). So that index.html becomes first index.html.1 and then
818 18.4 simultaneous parallel transfers
820 The client could be told to use maximum N simultaneous parallel transfers and
821 then just make sure that happens. It should of course not make more than one
822 connection to the same remote host. This would require the client to use the
823 multi interface. https://curl.haxx.se/bug/feature.cgi?id=1558595
825 Using the multi interface would also allow properly using parallel transfers
826 with HTTP/2 and supporting HTTP/2 server push from the command line.
828 18.5 provide formpost headers
830 Extending the capabilities of the multipart formposting. How about leaving
831 the ';type=foo' syntax as it is and adding an extra tag (headers) which
832 works like this: curl -F "coolfiles=@fil1.txt;headers=@fil1.hdr" where
833 fil1.hdr contains extra headers like
835 Content-Type: text/plain; charset=KOI8-R"
836 Content-Transfer-Encoding: base64
837 X-User-Comment: Please don't use browser specific HTML code
839 which should overwrite the program reasonable defaults (plain/text,
842 18.6 warning when setting an option
844 Display a warning when libcurl returns an error when setting an option.
845 This can be useful to tell when support for a particular feature hasn't been
846 compiled into the library.
848 18.7 warning when sending binary output to terminal
850 Provide a way that prompts the user for confirmation before binary data is
851 sent to the terminal, much in the style 'less' does it.
853 18.8 offer color-coded HTTP header output
855 By offering different color output on the header name and the header
856 contents, they could be made more readable and thus help users working on
859 18.9 Choose the name of file in braces for complex URLs
861 When using braces to download a list of URLs and you use complicated names
862 in the list of alternatives, it could be handy to allow curl to use other
865 Consider a way to offer that. Possibly like
866 {partURL1:name1,partURL2:name2,partURL3:name3} where the name following the
867 colon is the output name.
869 See https://github.com/curl/curl/issues/221
871 18.10 improve how curl works in a windows console window
873 If you pull the scrollbar when transferring with curl in a Windows console
874 window, the transfer is interrupted and can get disconnected. This can
875 probably be improved. See https://github.com/curl/curl/issues/322
877 18.11 -w output to stderr
879 -w is quite useful, but not to those of us who use curl without -o or -O
880 (such as for scripting through a higher level language). It would be nice to
881 have an option that is exactly like -w but sends it to stderr
882 instead. Proposed name: --write-stderr. See
883 https://github.com/curl/curl/issues/613
885 18.12 keep running, read instructions from pipe/socket
887 Provide an option that makes curl not exit after the last URL (or even work
888 without a given URL), and then make it read instructions passed on a pipe or
889 over a socket to make further instructions so that a second subsequent curl
890 invoke can talk to the still running instance and ask for transfers to get
891 done, and thus maintain its connection pool, DNS cache and more.
893 18.13 support metalink in http headers
895 Curl has support for downloading a metalink xml file, processing it, and then
896 downloading the target of the metalink. This is done via the --metalink option.
897 It would be nice if metalink also supported downloading via metalink
898 information that is stored in HTTP headers (RFC 6249). Theoretically this could
899 also be supported with the --metalink option.
901 See https://tools.ietf.org/html/rfc6249
903 See also https://lists.gnu.org/archive/html/bug-wget/2015-06/msg00034.html for
904 an implematation of this in wget.
906 18.14 --fail without --location should treat 3xx as a failure
908 To allow a command line like this to detect a redirect and consider it a
911 curl -v --fail -O https://example.com/curl-7.48.0.tar.gz
913 ... --fail must treat 3xx responses as failures too. The least problematic
914 way to implement this is probably to add that new logic in the command line
915 tool only and not in the underlying CURLOPT_FAILONERROR logic.
922 Consider extending 'roffit' to produce decent ASCII output, and use that
923 instead of (g)nroff when building src/tool_hugehelp.c
929 Make our own version of stunnel for simple port forwarding to enable HTTPS
930 and FTP-SSL tests without the stunnel dependency, and it could allow us to
931 provide test tools built with either OpenSSL or GnuTLS
933 20.2 nicer lacking perl message
935 If perl wasn't found by the configure script, don't attempt to run the tests
936 but explain something nice why it doesn't.
938 20.3 more protocols supported
940 Extend the test suite to include more protocols. The telnet could just do FTP
941 or http operations (for which we have test servers).
943 20.4 more platforms supported
945 Make the test suite work on more platforms. OpenBSD and Mac OS. Remove
946 fork()s and it should become even more portable.
948 20.5 Add support for concurrent connections
950 Tests 836, 882 and 938 were designed to verify that separate connections aren't
951 used when using different login credentials in protocols that shouldn't re-use
952 a connection under such circumstances.
954 Unfortunately, ftpserver.pl doesn't appear to support multiple concurrent
955 connections. The read while() loop seems to loop until it receives a disconnect
956 from the client, where it then enters the waiting for connections loop. When
957 the client opens a second connection to the server, the first connection hasn't
958 been dropped (unless it has been forced - which we shouldn't do in these tests)
959 and thus the wait for connections loop is never entered to receive the second
962 20.6 Use the RFC6265 test suite
964 A test suite made for HTTP cookies (RFC 6265) by Adam Barth is available at
965 https://github.com/abarth/http-state/tree/master/tests
967 It'd be really awesome if someone would write a script/setup that would run
968 curl with that test suite and detect deviances. Ideally, that would even be
969 incorporated into our regular test suite.
974 21.1 http-style HEAD output for FTP
976 #undef CURL_FTP_HTTPSTYLE_HEAD in lib/ftp.c to remove the HTTP-style headers
977 from being output in NOBODY requests over FTP
979 21.2 combine error codes
981 Combine some of the error codes to remove duplicates. The original
982 numbering should not be changed, and the old identifiers would be
983 macroed to the new ones in an CURL_NO_OLDIES section to help with
984 backward compatibility.
986 Candidates for removal and their replacements:
988 CURLE_FILE_COULDNT_READ_FILE => CURLE_REMOTE_FILE_NOT_FOUND
990 CURLE_FTP_COULDNT_RETR_FILE => CURLE_REMOTE_FILE_NOT_FOUND
992 CURLE_FTP_COULDNT_USE_REST => CURLE_RANGE_ERROR
994 CURLE_FUNCTION_NOT_FOUND => CURLE_FAILED_INIT
996 CURLE_LDAP_INVALID_URL => CURLE_URL_MALFORMAT
998 CURLE_TFTP_NOSUCHUSER => CURLE_TFTP_ILLEGAL
1000 CURLE_TFTP_NOTFOUND => CURLE_REMOTE_FILE_NOT_FOUND
1002 CURLE_TFTP_PERM => CURLE_REMOTE_ACCESS_DENIED
1004 21.3 extend CURLOPT_SOCKOPTFUNCTION prototype
1006 The current prototype only provides 'purpose' that tells what the
1007 connection/socket is for, but not any protocol or similar. It makes it hard
1008 for applications to differentiate on TCP vs UDP and even HTTP vs FTP and
1011 22. Next major release
1013 22.1 cleanup return codes
1015 curl_easy_cleanup() returns void, but curl_multi_cleanup() returns a
1016 CURLMcode. These should be changed to be the same.
1018 22.2 remove obsolete defines
1020 remove obsolete defines from curl/curl.h
1024 make several functions use size_t instead of int in their APIs
1026 22.4 remove several functions
1028 remove the following functions from the public API:
1032 curl_mprintf (and variations)
1038 They will instead become curlx_ - alternatives. That makes the curl app
1039 still capable of using them, by building with them from source.
1041 These functions have no purpose anymore:
1045 curl_multi_socket_all
1047 22.5 remove CURLOPT_FAILONERROR
1049 Remove support for CURLOPT_FAILONERROR, it has gotten too kludgy and weird
1050 internally. Let the app judge success or not for itself.
1052 22.6 remove CURLOPT_DNS_USE_GLOBAL_CACHE
1054 Remove support for a global DNS cache. Anything global is silly, and we
1055 already offer the share interface for the same functionality but done
1058 22.7 remove progress meter from libcurl
1060 The internally provided progress meter output doesn't belong in the library.
1061 Basically no application wants it (apart from curl) but instead applications
1062 can and should do their own progress meters using the progress callback.
1064 The progress callback should then be bumped as well to get proper 64bit
1065 variable types passed to it instead of doubles so that big files work
1068 22.8 remove 'curl_httppost' from public
1070 curl_formadd() was made to fill in a public struct, but the fact that the
1071 struct is public is never really used by application for their own advantage
1072 but instead often restricts how the form functions can or can't be modified.
1074 Changing them to return a private handle will benefit the implementation and
1075 allow us much greater freedoms while still maintaining a solid API and ABI.