Debian Codegraph @

deps/curl/docs/TODO

raw
1 _ _ ____ _
2 ___| | | | _ \| |
3 / __| | | | |_) | |
4 | (__| |_| | _ <| |___
5 \___|\___/|_| \_\_____|
6
7 Things that could be nice to do in the future
8
9 Things to do in project curl. Please tell us what you think, contribute and
10 send us patches that improve things!
11
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!
16
17 All bugs documented in the KNOWN_BUGS document are subject for fixing!
18
19 1. libcurl
20 1.1 TFO support on Windows
21 1.2 Consult %APPDATA% also for .netrc
22 1.3 struct lifreq
23 1.4 alt-svc sharing
24 1.5 get rid of PATH_MAX
25 1.7 Support HTTP/2 for HTTP(S) proxies
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 1.12 updated DNS server while running
31 1.13 c-ares and CURLOPT_OPENSOCKETFUNCTION
32 1.14 Typesafe curl_easy_setopt()
33 1.15 Monitor connections in the connection pool
34 1.16 Try to URL encode given URL
35 1.17 Add support for IRIs
36 1.18 try next proxy if one doesn't work
37 1.20 SRV and URI DNS records
38 1.22 CURLINFO_PAUSE_STATE
39 1.23 Offer API to flush the connection pool
40 1.24 TCP Fast Open for windows
41 1.25 Expose tried IP addresses that failed
42 1.27 hardcode the "localhost" addresses
43 1.28 FD_CLOEXEC
44 1.29 Upgrade to websockets
45 1.30 config file parsing
46
47 2. libcurl - multi interface
48 2.1 More non-blocking
49 2.2 Better support for same name resolves
50 2.3 Non-blocking curl_multi_remove_handle()
51 2.4 Split connect and authentication process
52 2.5 Edge-triggered sockets should work
53 2.6 multi upkeep
54
55 3. Documentation
56 3.2 Provide cmake config-file
57
58 4. FTP
59 4.1 HOST
60 4.2 Alter passive/active on failure and retry
61 4.3 Earlier bad letter detection
62 4.5 ASCII support
63 4.6 GSSAPI via Windows SSPI
64 4.7 STAT for LIST without data connection
65 4.8 Option to ignore private IP addresses in PASV response
66
67 5. HTTP
68 5.1 Better persistency for HTTP 1.0
69 5.3 Rearrange request header order
70 5.4 Allow SAN names in HTTP/2 server push
71 5.5 auth= in URLs
72
73 6. TELNET
74 6.1 ditch stdin
75 6.2 ditch telnet-specific select
76 6.3 feature negotiation debug data
77
78 7. SMTP
79 7.2 Enhanced capability support
80 7.3 Add CURLOPT_MAIL_CLIENT option
81
82 8. POP3
83 8.2 Enhanced capability support
84
85 9. IMAP
86 9.1 Enhanced capability support
87
88 10. LDAP
89 10.1 SASL based authentication mechanisms
90 10.2 CURLOPT_SSL_CTX_FUNCTION for LDAPS
91 10.3 Paged searches on LDAP server
92
93 11. SMB
94 11.1 File listing support
95 11.2 Honor file timestamps
96 11.3 Use NTLMv2
97 11.4 Create remote directories
98
99 12. New protocols
100
101 13. SSL
102 13.2 Provide mutex locking API
103 13.3 Support in-memory certs/ca certs/keys
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
108 13.8 Support DANE
109 13.10 Support Authority Information Access certificate extension (AIA)
110 13.11 Support intermediate & root pinning for PINNEDPUBLICKEY
111 13.12 Support HSTS
112 13.14 Support the clienthello extension
113
114 14. GnuTLS
115 14.2 check connection
116
117 15. WinSSL/SChannel
118 15.1 Add support for client certificate authentication
119 15.3 Add support for the --ciphers option
120 15.4 Add option to disable client certificate auto-send
121
122 16. SASL
123 16.1 Other authentication mechanisms
124 16.2 Add QOP support to GSSAPI authentication
125 16.3 Support binary messages (i.e.: non-base64)
126
127 17. SSH protocols
128 17.1 Multiplexing
129 17.2 Handle growing SFTP files
130 17.3 Support better than MD5 hostkey hash
131 17.4 Support CURLOPT_PREQUOTE
132
133 18. Command line tool
134 18.1 sync
135 18.2 glob posts
136 18.3 prevent file overwriting
137 18.4 --proxycommand
138 18.5 UTF-8 filenames in Content-Disposition
139 18.7 at least N milliseconds between requests
140 18.9 Choose the name of file in braces for complex URLs
141 18.10 improve how curl works in a windows console window
142 18.11 Windows: set attribute 'archive' for completed downloads
143 18.12 keep running, read instructions from pipe/socket
144 18.15 --retry should resume
145 18.16 send only part of --data
146 18.17 consider file name from the redirected URL with -O ?
147 18.18 retry on network is unreachable
148 18.19 expand ~/ in config files
149 18.20 host name sections in config files
150
151 19. Build
152 19.1 roffit
153 19.2 Enable PIE and RELRO by default
154 19.3 cmake test suite improvements
155
156 20. Test suite
157 20.1 SSL tunnel
158 20.2 nicer lacking perl message
159 20.3 more protocols supported
160 20.4 more platforms supported
161 20.5 Add support for concurrent connections
162 20.6 Use the RFC6265 test suite
163 20.7 Support LD_PRELOAD on macOS
164 20.8 Run web-platform-tests url tests
165
166 21. Next SONAME bump
167 21.1 http-style HEAD output for FTP
168 21.2 combine error codes
169 21.3 extend CURLOPT_SOCKOPTFUNCTION prototype
170
171 22. Next major release
172 22.1 cleanup return codes
173 22.2 remove obsolete defines
174 22.3 size_t
175 22.4 remove several functions
176 22.5 remove CURLOPT_FAILONERROR
177 22.7 remove progress meter from libcurl
178 22.8 remove 'curl_httppost' from public
179
180 ==============================================================================
181
182 1. libcurl
183
184 1.1 TFO support on Windows
185
186 TCP Fast Open is supported on several platforms but not on Windows. Work on
187 this was once started but never finished.
188
189 See https://github.com/curl/curl/pull/3378
190
191 1.2 Consult %APPDATA% also for .netrc
192
193 %APPDATA%\.netrc is not considered when running on Windows. Shouldn't it?
194
195 See https://github.com/curl/curl/issues/4016
196
197 1.3 struct lifreq
198
199 Use 'struct lifreq' and SIOCGLIFADDR instead of 'struct ifreq' and
200 SIOCGIFADDR on newer Solaris versions as they claim the latter is obsolete.
201 To support IPv6 interface addresses for network interfaces properly.
202
203 1.4 alt-svc sharing
204
205 The share interface could benefit from allowing the alt-svc cache to be
206 possible to share between easy handles.
207
208 See https://github.com/curl/curl/issues/4476
209
210 1.5 get rid of PATH_MAX
211
212 Having code use and rely on PATH_MAX is not nice:
213 https://insanecoding.blogspot.com/2007/11/pathmax-simply-isnt.html
214
215 Currently the libssh2 SSH based code uses it, but to remove PATH_MAX from
216 there we need libssh2 to properly tell us when we pass in a too small buffer
217 and its current API (as of libssh2 1.2.7) doesn't.
218
219 1.7 Support HTTP/2 for HTTP(S) proxies
220
221 Support for doing HTTP/2 to HTTP and HTTPS proxies is still missing.
222
223 See https://github.com/curl/curl/issues/3570
224
225 1.8 CURLOPT_RESOLVE for any port number
226
227 This option allows applications to set a replacement IP address for a given
228 host + port pair. Consider making support for providing a replacement address
229 for the host name on all port numbers.
230
231 See https://github.com/curl/curl/issues/1264
232
233 1.9 Cache negative name resolves
234
235 A name resolve that has failed is likely to fail when made again within a
236 short period of time. Currently we only cache positive responses.
237
238 1.10 auto-detect proxy
239
240 libcurl could be made to detect the system proxy setup automatically and use
241 that. On Windows, macOS and Linux desktops for example.
242
243 The pull-request to use libproxy for this was deferred due to doubts on the
244 reliability of the dependency and how to use it:
245 https://github.com/curl/curl/pull/977
246
247 libdetectproxy is a (C++) library for detecting the proxy on Windows
248 https://github.com/paulharris/libdetectproxy
249
250 1.11 minimize dependencies with dynamically loaded modules
251
252 We can create a system with loadable modules/plug-ins, where these modules
253 would be the ones that link to 3rd party libs. That would allow us to avoid
254 having to load ALL dependencies since only the necessary ones for this
255 app/invoke/used protocols would be necessary to load. See
256 https://github.com/curl/curl/issues/349
257
258 1.12 updated DNS server while running
259
260 If /etc/resolv.conf gets updated while a program using libcurl is running, it
261 is may cause name resolves to fail unless res_init() is called. We should
262 consider calling res_init() + retry once unconditionally on all name resolve
263 failures to mitigate against this. Firefox works like that. Note that Windows
264 doesn't have res_init() or an alternative.
265
266 https://github.com/curl/curl/issues/2251
267
268 1.13 c-ares and CURLOPT_OPENSOCKETFUNCTION
269
270 curl will create most sockets via the CURLOPT_OPENSOCKETFUNCTION callback and
271 close them with the CURLOPT_CLOSESOCKETFUNCTION callback. However, c-ares
272 does not use those functions and instead opens and closes the sockets
273 itself. This means that when curl passes the c-ares socket to the
274 CURLMOPT_SOCKETFUNCTION it isn't owned by the application like other sockets.
275
276 See https://github.com/curl/curl/issues/2734
277
278 1.14 Typesafe curl_easy_setopt()
279
280 One of the most common problems in libcurl using applications is the lack of
281 type checks for curl_easy_setopt() which happens because it accepts varargs
282 and thus can take any type.
283
284 One possible solution to this is to introduce a few different versions of the
285 setopt version for the different kinds of data you can set.
286
287 curl_easy_set_num() - sets a long value
288
289 curl_easy_set_large() - sets a curl_off_t value
290
291 curl_easy_set_ptr() - sets a pointer
292
293 curl_easy_set_cb() - sets a callback PLUS its callback data
294
295 1.15 Monitor connections in the connection pool
296
297 libcurl's connection cache or pool holds a number of open connections for the
298 purpose of possible subsequent connection reuse. It may contain a few up to a
299 significant amount of connections. Currently, libcurl leaves all connections
300 as they are and first when a connection is iterated over for matching or
301 reuse purpose it is verified that it is still alive.
302
303 Those connections may get closed by the server side for idleness or they may
304 get a HTTP/2 ping from the peer to verify that they're still alive. By adding
305 monitoring of the connections while in the pool, libcurl can detect dead
306 connections (and close them) better and earlier, and it can handle HTTP/2
307 pings to keep such ones alive even when not actively doing transfers on them.
308
309 1.16 Try to URL encode given URL
310
311 Given a URL that for example contains spaces, libcurl could have an option
312 that would try somewhat harder than it does now and convert spaces to %20 and
313 perhaps URL encoded byte values over 128 etc (basically do what the redirect
314 following code already does).
315
316 https://github.com/curl/curl/issues/514
317
318 1.17 Add support for IRIs
319
320 IRIs (RFC 3987) allow localized, non-ascii, names in the URL. To properly
321 support this, curl/libcurl would need to translate/encode the given input
322 from the input string encoding into percent encoded output "over the wire".
323
324 To make that work smoothly for curl users even on Windows, curl would
325 probably need to be able to convert from several input encodings.
326
327 1.18 try next proxy if one doesn't work
328
329 Allow an application to specify a list of proxies to try, and failing to
330 connect to the first go on and try the next instead until the list is
331 exhausted. Browsers support this feature at least when they specify proxies
332 using PACs.
333
334 https://github.com/curl/curl/issues/896
335
336 1.20 SRV and URI DNS records
337
338 Offer support for resolving SRV and URI DNS records for libcurl to know which
339 server to connect to for various protocols (including HTTP!).
340
341 1.22 CURLINFO_PAUSE_STATE
342
343 Return information about the transfer's current pause state, in both
344 directions. https://github.com/curl/curl/issues/2588
345
346 1.23 Offer API to flush the connection pool
347
348 Sometimes applications want to flush all the existing connections kept alive.
349 An API could allow a forced flush or just a forced loop that would properly
350 close all connections that have been closed by the server already.
351
352 1.24 TCP Fast Open for windows
353
354 libcurl supports the CURLOPT_TCP_FASTOPEN option since 7.49.0 for Linux and
355 Mac OS. Windows supports TCP Fast Open starting with Windows 10, version 1607
356 and we should add support for it.
357
358 1.25 Expose tried IP addresses that failed
359
360 When libcurl fails to connect to a host, it should be able to offer the
361 application the list of IP addresses that were used in the attempt.
362
363 https://github.com/curl/curl/issues/2126
364
365 1.27 hardcode the "localhost" addresses
366
367 There's this new spec getting adopted that says "localhost" should always and
368 unconditionally be a local address and not get resolved by a DNS server. A
369 fine way for curl to fix this would be to simply hard-code the response to
370 127.0.0.1 and/or ::1 (depending on what IP versions that are requested). This
371 is what the browsers probably will do with this hostname.
372
373 https://bugzilla.mozilla.org/show_bug.cgi?id=1220810
374
375 https://tools.ietf.org/html/draft-ietf-dnsop-let-localhost-be-localhost-02
376
377 1.28 FD_CLOEXEC
378
379 It sets the close-on-exec flag for the file descriptor, which causes the file
380 descriptor to be automatically (and atomically) closed when any of the
381 exec-family functions succeed. Should probably be set by default?
382
383 https://github.com/curl/curl/issues/2252
384
385 1.29 Upgrade to websockets
386
387 libcurl could offer a smoother path to get to a websocket connection.
388 See https://github.com/curl/curl/issues/3523
389
390 Michael Kaufmann suggestion here:
391 https://curl.haxx.se/video/curlup-2017/2017-03-19_05_Michael_Kaufmann_Websocket_support_for_curl.mp4
392
393 1.30 config file parsing
394
395 Consider providing an API, possibly in a separate companion library, for
396 parsing a config file like curl's -K/--config option to allow applications to
397 get the same ability to read curl options from files.
398
399 See https://github.com/curl/curl/issues/3698
400
401 2. libcurl - multi interface
402
403 2.1 More non-blocking
404
405 Make sure we don't ever loop because of non-blocking sockets returning
406 EWOULDBLOCK or similar. Blocking cases include:
407
408 - Name resolves on non-windows unless c-ares or the threaded resolver is used
409 - file:// transfers
410 - TELNET transfers
411 - The "DONE" operation (post transfer protocol-specific actions) for the
412 protocols SFTP, SMTP, FTP. Fixing Curl_done() for this is a worthy task.
413
414 2.2 Better support for same name resolves
415
416 If a name resolve has been initiated for name NN and a second easy handle
417 wants to resolve that name as well, make it wait for the first resolve to end
418 up in the cache instead of doing a second separate resolve. This is
419 especially needed when adding many simultaneous handles using the same host
420 name when the DNS resolver can get flooded.
421
422 2.3 Non-blocking curl_multi_remove_handle()
423
424 The multi interface has a few API calls that assume a blocking behavior, like
425 add_handle() and remove_handle() which limits what we can do internally. The
426 multi API need to be moved even more into a single function that "drives"
427 everything in a non-blocking manner and signals when something is done. A
428 remove or add would then only ask for the action to get started and then
429 multi_perform() etc still be called until the add/remove is completed.
430
431 2.4 Split connect and authentication process
432
433 The multi interface treats the authentication process as part of the connect
434 phase. As such any failures during authentication won't trigger the relevant
435 QUIT or LOGOFF for protocols such as IMAP, POP3 and SMTP.
436
437 2.5 Edge-triggered sockets should work
438
439 The multi_socket API should work with edge-triggered socket events. One of
440 the internal actions that need to be improved for this to work perfectly is
441 the 'maxloops' handling in transfer.c:readwrite_data().
442
443 2.6 multi upkeep
444
445 In libcurl 7.62.0 we introduced curl_easy_upkeep. It unfortunately only works
446 on easy handles. We should introduces a version of that for the multi handle,
447 and also consider doing "upkeep" automatically on connections in the
448 connection pool when the multi handle is in used.
449
450 See https://github.com/curl/curl/issues/3199
451
452 3. Documentation
453
454 3.2 Provide cmake config-file
455
456 A config-file package is a set of files provided by us to allow applications
457 to write cmake scripts to find and use libcurl easier. See
458 https://github.com/curl/curl/issues/885
459
460 4. FTP
461
462 4.1 HOST
463
464 HOST is a command for a client to tell which host name to use, to offer FTP
465 servers named-based virtual hosting:
466
467 https://tools.ietf.org/html/rfc7151
468
469 4.2 Alter passive/active on failure and retry
470
471 When trying to connect passively to a server which only supports active
472 connections, libcurl returns CURLE_FTP_WEIRD_PASV_REPLY and closes the
473 connection. There could be a way to fallback to an active connection (and
474 vice versa). https://curl.haxx.se/bug/feature.cgi?id=1754793
475
476 4.3 Earlier bad letter detection
477
478 Make the detection of (bad) %0d and %0a codes in FTP URL parts earlier in the
479 process to avoid doing a resolve and connect in vain.
480
481 4.5 ASCII support
482
483 FTP ASCII transfers do not follow RFC959. They don't convert the data
484 accordingly.
485
486 4.6 GSSAPI via Windows SSPI
487
488 In addition to currently supporting the SASL GSSAPI mechanism (Kerberos V5)
489 via third-party GSS-API libraries, such as Heimdal or MIT Kerberos, also add
490 support for GSSAPI authentication via Windows SSPI.
491
492 4.7 STAT for LIST without data connection
493
494 Some FTP servers allow STAT for listing directories instead of using LIST,
495 and the response is then sent over the control connection instead of as the
496 otherwise usedw data connection: https://www.nsftools.com/tips/RawFTP.htm#STAT
497
498 This is not detailed in any FTP specification.
499
500 4.8 Option to ignore private IP addresses in PASV response
501
502 Some servers respond with and some other FTP client implementations can
503 ignore private (RFC 1918 style) IP addresses when received in PASV responses.
504 To consider for libcurl as well. See https://github.com/curl/curl/issues/1455
505
506 5. HTTP
507
508 5.1 Better persistency for HTTP 1.0
509
510 "Better" support for persistent connections over HTTP 1.0
511 https://curl.haxx.se/bug/feature.cgi?id=1089001
512
513 5.3 Rearrange request header order
514
515 Server implementors often make an effort to detect browser and to reject
516 clients it can detect to not match. One of the last details we cannot yet
517 control in libcurl's HTTP requests, which also can be exploited to detect
518 that libcurl is in fact used even when it tries to impersonate a browser, is
519 the order of the request headers. I propose that we introduce a new option in
520 which you give headers a value, and then when the HTTP request is built it
521 sorts the headers based on that number. We could then have internally created
522 headers use a default value so only headers that need to be moved have to be
523 specified.
524
525 5.4 Allow SAN names in HTTP/2 server push
526
527 curl only allows HTTP/2 push promise if the provided :authority header value
528 exactly matches the host name given in the URL. It could be extended to allow
529 any name that would match the Subject Alternative Names in the server's TLS
530 certificate.
531
532 See https://github.com/curl/curl/pull/3581
533
534 5.5 auth= in URLs
535
536 Add the ability to specify the preferred authentication mechanism to use by
537 using ;auth=<mech> in the login part of the URL.
538
539 For example:
540
541 http://test:pass;[email protected] would be equivalent to specifying
542 --user test:pass;auth=NTLM or --user test:pass --ntlm from the command line.
543
544 Additionally this should be implemented for proxy base URLs as well.
545
546
547 6. TELNET
548
549 6.1 ditch stdin
550
551 Reading input (to send to the remote server) on stdin is a crappy solution
552 for library purposes. We need to invent a good way for the application to be
553 able to provide the data to send.
554
555 6.2 ditch telnet-specific select
556
557 Move the telnet support's network select() loop go away and merge the code
558 into the main transfer loop. Until this is done, the multi interface won't
559 work for telnet.
560
561 6.3 feature negotiation debug data
562
563 Add telnet feature negotiation data to the debug callback as header data.
564
565
566 7. SMTP
567
568 7.2 Enhanced capability support
569
570 Add the ability, for an application that uses libcurl, to obtain the list of
571 capabilities returned from the EHLO command.
572
573 7.3 Add CURLOPT_MAIL_CLIENT option
574
575 Rather than use the URL to specify the mail client string to present in the
576 HELO and EHLO commands, libcurl should support a new CURLOPT specifically for
577 specifying this data as the URL is non-standard and to be honest a bit of a
578 hack ;-)
579
580 Please see the following thread for more information:
581 https://curl.haxx.se/mail/lib-2012-05/0178.html
582
583
584 8. POP3
585
586 8.2 Enhanced capability support
587
588 Add the ability, for an application that uses libcurl, to obtain the list of
589 capabilities returned from the CAPA command.
590
591 9. IMAP
592
593 9.1 Enhanced capability support
594
595 Add the ability, for an application that uses libcurl, to obtain the list of
596 capabilities returned from the CAPABILITY command.
597
598 10. LDAP
599
600 10.1 SASL based authentication mechanisms
601
602 Currently the LDAP module only supports ldap_simple_bind_s() in order to bind
603 to an LDAP server. However, this function sends username and password details
604 using the simple authentication mechanism (as clear text). However, it should
605 be possible to use ldap_bind_s() instead specifying the security context
606 information ourselves.
607
608 10.2 CURLOPT_SSL_CTX_FUNCTION for LDAPS
609
610 CURLOPT_SSL_CTX_FUNCTION works perfectly for HTTPS and email protocols, but
611 it has no effect for LDAPS connections.
612
613 https://github.com/curl/curl/issues/4108
614
615 10.3 Paged searches on LDAP server
616
617 https://github.com/curl/curl/issues/4452
618
619 11. SMB
620
621 11.1 File listing support
622
623 Add support for listing the contents of a SMB share. The output should probably
624 be the same as/similar to FTP.
625
626 11.2 Honor file timestamps
627
628 The timestamp of the transferred file should reflect that of the original file.
629
630 11.3 Use NTLMv2
631
632 Currently the SMB authentication uses NTLMv1.
633
634 11.4 Create remote directories
635
636 Support for creating remote directories when uploading a file to a directory
637 that doesn't exist on the server, just like --ftp-create-dirs.
638
639 12. New protocols
640
641 13. SSL
642
643 13.2 Provide mutex locking API
644
645 Provide a libcurl API for setting mutex callbacks in the underlying SSL
646 library, so that the same application code can use mutex-locking
647 independently of OpenSSL or GnutTLS being used.
648
649 13.3 Support in-memory certs/ca certs/keys
650
651 You can specify the private and public keys for SSH/SSL as file paths. Some
652 programs want to avoid using files and instead just pass them as in-memory
653 data blobs. There's probably a challenge to make this work across the
654 plethory of different TLS and SSH backends that curl supports.
655 https://github.com/curl/curl/issues/2310
656
657 13.4 Cache/share OpenSSL contexts
658
659 "Look at SSL cafile - quick traces look to me like these are done on every
660 request as well, when they should only be necessary once per SSL context (or
661 once per handle)". The major improvement we can rather easily do is to make
662 sure we don't create and kill a new SSL "context" for every request, but
663 instead make one for every connection and re-use that SSL context in the same
664 style connections are re-used. It will make us use slightly more memory but
665 it will libcurl do less creations and deletions of SSL contexts.
666
667 Technically, the "caching" is probably best implemented by getting added to
668 the share interface so that easy handles who want to and can reuse the
669 context specify that by sharing with the right properties set.
670
671 https://github.com/curl/curl/issues/1110
672
673 13.5 Export session ids
674
675 Add an interface to libcurl that enables "session IDs" to get
676 exported/imported. Cris Bailiff said: "OpenSSL has functions which can
677 serialise the current SSL state to a buffer of your choice, and recover/reset
678 the state from such a buffer at a later date - this is used by mod_ssl for
679 apache to implement and SSL session ID cache".
680
681 13.6 Provide callback for cert verification
682
683 OpenSSL supports a callback for customised verification of the peer
684 certificate, but this doesn't seem to be exposed in the libcurl APIs. Could
685 it be? There's so much that could be done if it were!
686
687 13.7 improve configure --with-ssl
688
689 make the configure --with-ssl option first check for OpenSSL, then GnuTLS,
690 then NSS...
691
692 13.8 Support DANE
693
694 DNS-Based Authentication of Named Entities (DANE) is a way to provide SSL
695 keys and certs over DNS using DNSSEC as an alternative to the CA model.
696 https://www.rfc-editor.org/rfc/rfc6698.txt
697
698 An initial patch was posted by Suresh Krishnaswamy on March 7th 2013
699 (https://curl.haxx.se/mail/lib-2013-03/0075.html) but it was a too simple
700 approach. See Daniel's comments:
701 https://curl.haxx.se/mail/lib-2013-03/0103.html . libunbound may be the
702 correct library to base this development on.
703
704 Björn Stenberg wrote a separate initial take on DANE that was never
705 completed.
706
707 13.10 Support Authority Information Access certificate extension (AIA)
708
709 AIA can provide various things like CRLs but more importantly information
710 about intermediate CA certificates that can allow validation path to be
711 fulfilled when the HTTPS server doesn't itself provide them.
712
713 Since AIA is about downloading certs on demand to complete a TLS handshake,
714 it is probably a bit tricky to get done right.
715
716 See https://github.com/curl/curl/issues/2793
717
718 13.11 Support intermediate & root pinning for PINNEDPUBLICKEY
719
720 CURLOPT_PINNEDPUBLICKEY does not consider the hashes of intermediate & root
721 certificates when comparing the pinned keys. Therefore it is not compatible
722 with "HTTP Public Key Pinning" as there also intermediate and root certificates
723 can be pinned. This is very useful as it prevents webadmins from "locking
724 themself out of their servers".
725
726 Adding this feature would make curls pinning 100% compatible to HPKP and allow
727 more flexible pinning.
728
729 13.12 Support HSTS
730
731 "HTTP Strict Transport Security" is TOFU (trust on first use), time-based
732 features indicated by a HTTP header send by the webserver. It is widely used
733 in browsers and it's purpose is to prevent insecure HTTP connections after
734 a previous HTTPS connection. It protects against SSLStripping attacks.
735
736 Doc: https://developer.mozilla.org/en-US/docs/Web/Security/HTTP_strict_transport_security
737 RFC 6797: https://tools.ietf.org/html/rfc6797
738
739 13.14 Support the clienthello extension
740
741 Certain stupid networks and middle boxes have a problem with SSL handshake
742 pakets that are within a certain size range because how that sets some bits
743 that previously (in older TLS version) were not set. The clienthello
744 extension adds padding to avoid that size range.
745
746 https://tools.ietf.org/html/rfc7685
747 https://github.com/curl/curl/issues/2299
748
749 14. GnuTLS
750
751 14.2 check connection
752
753 Add a way to check if the connection seems to be alive, to correspond to the
754 SSL_peak() way we use with OpenSSL.
755
756 15. WinSSL/SChannel
757
758 15.1 Add support for client certificate authentication
759
760 WinSSL/SChannel currently makes use of the OS-level system and user
761 certificate and private key stores. This does not allow the application
762 or the user to supply a custom client certificate using curl or libcurl.
763
764 Therefore support for the existing -E/--cert and --key options should be
765 implemented by supplying a custom certificate to the SChannel APIs, see:
766 - Getting a Certificate for Schannel
767 https://msdn.microsoft.com/en-us/library/windows/desktop/aa375447.aspx
768
769 15.3 Add support for the --ciphers option
770
771 The cipher suites used by WinSSL/SChannel are configured on an OS-level
772 instead of an application-level. This does not allow the application or
773 the user to customize the configured cipher suites using curl or libcurl.
774
775 Therefore support for the existing --ciphers option should be implemented
776 by mapping the OpenSSL/GnuTLS cipher suites to the SChannel APIs, see
777 - Specifying Schannel Ciphers and Cipher Strengths
778 https://msdn.microsoft.com/en-us/library/windows/desktop/aa380161.aspx
779
780 15.4 Add option to disable client certificate auto-send
781
782 Microsoft says "By default, Schannel will, with no notification to the client,
783 attempt to locate a client certificate and send it to the server." That could
784 be considered a privacy violation and unexpected.
785
786 Some Windows users have come to expect that default behavior and to change the
787 default to make it consistent with other SSL backends would be a breaking
788 change. An option should be added that can be used to disable the default
789 Schannel auto-send behavior.
790
791 https://github.com/curl/curl/issues/2262
792
793 16. SASL
794
795 16.1 Other authentication mechanisms
796
797 Add support for other authentication mechanisms such as OLP,
798 GSS-SPNEGO and others.
799
800 16.2 Add QOP support to GSSAPI authentication
801
802 Currently the GSSAPI authentication only supports the default QOP of auth
803 (Authentication), whilst Kerberos V5 supports both auth-int (Authentication
804 with integrity protection) and auth-conf (Authentication with integrity and
805 privacy protection).
806
807 16.3 Support binary messages (i.e.: non-base64)
808
809 Mandatory to support LDAP SASL authentication.
810
811
812 17. SSH protocols
813
814 17.1 Multiplexing
815
816 SSH is a perfectly fine multiplexed protocols which would allow libcurl to do
817 multiple parallel transfers from the same host using the same connection,
818 much in the same spirit as HTTP/2 does. libcurl however does not take
819 advantage of that ability but will instead always create a new connection for
820 new transfers even if an existing connection already exists to the host.
821
822 To fix this, libcurl would have to detect an existing connection and "attach"
823 the new transfer to the existing one.
824
825 17.2 Handle growing SFTP files
826
827 The SFTP code in libcurl checks the file size *before* a transfer starts and
828 then proceeds to transfer exactly that amount of data. If the remote file
829 grows while the transfer is in progress libcurl won't notice and will not
830 adapt. The OpenSSH SFTP command line tool does and libcurl could also just
831 attempt to download more to see if there is more to get...
832
833 https://github.com/curl/curl/issues/4344
834
835 17.3 Support better than MD5 hostkey hash
836
837 libcurl offers the CURLOPT_SSH_HOST_PUBLIC_KEY_MD5 option for verifying the
838 server's key. MD5 is generally being deprecated so we should implement
839 support for stronger hashing algorithms. libssh2 itself is what provides this
840 underlying functionality and it supports at least SHA-1 as an alternative.
841 SHA-1 is also being deprecated these days so we should consider working with
842 libssh2 to instead offer support for SHA-256 or similar.
843
844 17.4 Support CURLOPT_PREQUOTE
845
846 The two other QUOTE options are supported for SFTP, but this was left out for
847 unknown reasons!
848
849 18. Command line tool
850
851 18.1 sync
852
853 "curl --sync http://example.com/feed[1-100].rss" or
854 "curl --sync http://example.net/{index,calendar,history}.html"
855
856 Downloads a range or set of URLs using the remote name, but only if the
857 remote file is newer than the local file. A Last-Modified HTTP date header
858 should also be used to set the mod date on the downloaded file.
859
860 18.2 glob posts
861
862 Globbing support for -d and -F, as in 'curl -d "name=foo[0-9]" URL'.
863 This is easily scripted though.
864
865 18.3 prevent file overwriting
866
867 Add an option that prevents curl from overwriting existing local files. When
868 used, and there already is an existing file with the target file name
869 (either -O or -o), a number should be appended (and increased if already
870 existing). So that index.html becomes first index.html.1 and then
871 index.html.2 etc.
872
873 18.4 --proxycommand
874
875 Allow the user to make curl run a command and use its stdio to make requests
876 and not do any network connection by itself. Example:
877
878 curl --proxycommand 'ssh [email protected] -W 10.1.1.75 80' \
879 http://some/otherwise/unavailable/service.php
880
881 See https://github.com/curl/curl/issues/4941
882
883 18.5 UTF-8 filenames in Content-Disposition
884
885 RFC 6266 documents how UTF-8 names can be passed to a client in the
886 Content-Disposition header, and curl does not support this.
887
888 https://github.com/curl/curl/issues/1888
889
890 18.7 at least N milliseconds between requests
891
892 Allow curl command lines issue a lot of request against services that limit
893 users to no more than N requests/second or similar. Could be implemented with
894 an option asking that at least a certain time has elapsed since the previous
895 request before the next one will be performed. Example:
896
897 $ curl "https://example.com/api?input=[1-1000]" -d yadayada --after 500
898
899 See https://github.com/curl/curl/issues/3920
900
901 18.9 Choose the name of file in braces for complex URLs
902
903 When using braces to download a list of URLs and you use complicated names
904 in the list of alternatives, it could be handy to allow curl to use other
905 names when saving.
906
907 Consider a way to offer that. Possibly like
908 {partURL1:name1,partURL2:name2,partURL3:name3} where the name following the
909 colon is the output name.
910
911 See https://github.com/curl/curl/issues/221
912
913 18.10 improve how curl works in a windows console window
914
915 If you pull the scrollbar when transferring with curl in a Windows console
916 window, the transfer is interrupted and can get disconnected. This can
917 probably be improved. See https://github.com/curl/curl/issues/322
918
919 18.11 Windows: set attribute 'archive' for completed downloads
920
921 The archive bit (FILE_ATTRIBUTE_ARCHIVE, 0x20) separates files that shall be
922 backed up from those that are either not ready or have not changed.
923
924 Downloads in progress are neither ready to be backed up, nor should they be
925 opened by a different process. Only after a download has been completed it's
926 sensible to include it in any integer snapshot or backup of the system.
927
928 See https://github.com/curl/curl/issues/3354
929
930 18.12 keep running, read instructions from pipe/socket
931
932 Provide an option that makes curl not exit after the last URL (or even work
933 without a given URL), and then make it read instructions passed on a pipe or
934 over a socket to make further instructions so that a second subsequent curl
935 invoke can talk to the still running instance and ask for transfers to get
936 done, and thus maintain its connection pool, DNS cache and more.
937
938 18.15 --retry should resume
939
940 When --retry is used and curl actually retries transfer, it should use the
941 already transferred data and do a resumed transfer for the rest (when
942 possible) so that it doesn't have to transfer the same data again that was
943 already transferred before the retry.
944
945 See https://github.com/curl/curl/issues/1084
946
947 18.16 send only part of --data
948
949 When the user only wants to send a small piece of the data provided with
950 --data or --data-binary, like when that data is a huge file, consider a way
951 to specify that curl should only send a piece of that. One suggested syntax
952 would be: "--data-binary @largefile.zip!1073741823-2147483647".
953
954 See https://github.com/curl/curl/issues/1200
955
956 18.17 consider file name from the redirected URL with -O ?
957
958 When a user gives a URL and uses -O, and curl follows a redirect to a new
959 URL, the file name is not extracted and used from the newly redirected-to URL
960 even if the new URL may have a much more sensible file name.
961
962 This is clearly documented and helps for security since there's no surprise
963 to users which file name that might get overwritten. But maybe a new option
964 could allow for this or maybe -J should imply such a treatment as well as -J
965 already allows for the server to decide what file name to use so it already
966 provides the "may overwrite any file" risk.
967
968 This is extra tricky if the original URL has no file name part at all since
969 then the current code path will error out with an error message, and we can't
970 *know* already at that point if curl will be redirected to a URL that has a
971 file name...
972
973 See https://github.com/curl/curl/issues/1241
974
975 18.18 retry on network is unreachable
976
977 The --retry option retries transfers on "transient failures". We later added
978 --retry-connrefused to also retry for "connection refused" errors.
979
980 Suggestions have been brought to also allow retry on "network is unreachable"
981 errors and while totally reasonable, maybe we should consider a way to make
982 this more configurable than to add a new option for every new error people
983 want to retry for?
984
985 https://github.com/curl/curl/issues/1603
986
987 18.19 expand ~/ in config files
988
989 For example .curlrc could benefit from being able to do this.
990
991 See https://github.com/curl/curl/issues/2317
992
993 18.20 host name sections in config files
994
995 config files would be more powerful if they could set different
996 configurations depending on used URLs, host name or possibly origin. Then a
997 default .curlrc could a specific user-agent only when doing requests against
998 a certain site.
999
1000
1001 19. Build
1002
1003 19.1 roffit
1004
1005 Consider extending 'roffit' to produce decent ASCII output, and use that
1006 instead of (g)nroff when building src/tool_hugehelp.c
1007
1008 19.2 Enable PIE and RELRO by default
1009
1010 Especially when having programs that execute curl via the command line, PIE
1011 renders the exploitation of memory corruption vulnerabilities a lot more
1012 difficult. This can be attributed to the additional information leaks being
1013 required to conduct a successful attack. RELRO, on the other hand, masks
1014 different binary sections like the GOT as read-only and thus kills a handful
1015 of techniques that come in handy when attackers are able to arbitrarily
1016 overwrite memory. A few tests showed that enabling these features had close
1017 to no impact, neither on the performance nor on the general functionality of
1018 curl.
1019
1020 19.3 cmake test suite improvements
1021
1022 The cmake build doesn't support 'make show' so it doesn't know which tests
1023 are in the makefile or not (making appveyor builds do many false warnings
1024 about it) nor does it support running the test suite if building out-of-tree.
1025
1026 See https://github.com/curl/curl/issues/3109
1027
1028 20. Test suite
1029
1030 20.1 SSL tunnel
1031
1032 Make our own version of stunnel for simple port forwarding to enable HTTPS
1033 and FTP-SSL tests without the stunnel dependency, and it could allow us to
1034 provide test tools built with either OpenSSL or GnuTLS
1035
1036 20.2 nicer lacking perl message
1037
1038 If perl wasn't found by the configure script, don't attempt to run the tests
1039 but explain something nice why it doesn't.
1040
1041 20.3 more protocols supported
1042
1043 Extend the test suite to include more protocols. The telnet could just do FTP
1044 or http operations (for which we have test servers).
1045
1046 20.4 more platforms supported
1047
1048 Make the test suite work on more platforms. OpenBSD and Mac OS. Remove
1049 fork()s and it should become even more portable.
1050
1051 20.5 Add support for concurrent connections
1052
1053 Tests 836, 882 and 938 were designed to verify that separate connections
1054 aren't used when using different login credentials in protocols that
1055 shouldn't re-use a connection under such circumstances.
1056
1057 Unfortunately, ftpserver.pl doesn't appear to support multiple concurrent
1058 connections. The read while() loop seems to loop until it receives a
1059 disconnect from the client, where it then enters the waiting for connections
1060 loop. When the client opens a second connection to the server, the first
1061 connection hasn't been dropped (unless it has been forced - which we
1062 shouldn't do in these tests) and thus the wait for connections loop is never
1063 entered to receive the second connection.
1064
1065 20.6 Use the RFC6265 test suite
1066
1067 A test suite made for HTTP cookies (RFC 6265) by Adam Barth is available at
1068 https://github.com/abarth/http-state/tree/master/tests
1069
1070 It'd be really awesome if someone would write a script/setup that would run
1071 curl with that test suite and detect deviances. Ideally, that would even be
1072 incorporated into our regular test suite.
1073
1074 20.7 Support LD_PRELOAD on macOS
1075
1076 LD_RELOAD doesn't work on macOS, but there are tests which require it to run
1077 properly. Look into making the preload support in runtests.pl portable such
1078 that it uses DYLD_INSERT_LIBRARIES on macOS.
1079
1080 20.8 Run web-platform-tests url tests
1081
1082 Run web-platform-tests url tests and compare results with browsers on wpt.fyi
1083
1084 It would help us find issues to fix and help us document where our parser
1085 differs from the WHATWG URL spec parsers.
1086
1087 See https://github.com/curl/curl/issues/4477
1088
1089 21. Next SONAME bump
1090
1091 21.1 http-style HEAD output for FTP
1092
1093 #undef CURL_FTP_HTTPSTYLE_HEAD in lib/ftp.c to remove the HTTP-style headers
1094 from being output in NOBODY requests over FTP
1095
1096 21.2 combine error codes
1097
1098 Combine some of the error codes to remove duplicates. The original
1099 numbering should not be changed, and the old identifiers would be
1100 macroed to the new ones in an CURL_NO_OLDIES section to help with
1101 backward compatibility.
1102
1103 Candidates for removal and their replacements:
1104
1105 CURLE_FILE_COULDNT_READ_FILE => CURLE_REMOTE_FILE_NOT_FOUND
1106
1107 CURLE_FTP_COULDNT_RETR_FILE => CURLE_REMOTE_FILE_NOT_FOUND
1108
1109 CURLE_FTP_COULDNT_USE_REST => CURLE_RANGE_ERROR
1110
1111 CURLE_FUNCTION_NOT_FOUND => CURLE_FAILED_INIT
1112
1113 CURLE_LDAP_INVALID_URL => CURLE_URL_MALFORMAT
1114
1115 CURLE_TFTP_NOSUCHUSER => CURLE_TFTP_ILLEGAL
1116
1117 CURLE_TFTP_NOTFOUND => CURLE_REMOTE_FILE_NOT_FOUND
1118
1119 CURLE_TFTP_PERM => CURLE_REMOTE_ACCESS_DENIED
1120
1121 21.3 extend CURLOPT_SOCKOPTFUNCTION prototype
1122
1123 The current prototype only provides 'purpose' that tells what the
1124 connection/socket is for, but not any protocol or similar. It makes it hard
1125 for applications to differentiate on TCP vs UDP and even HTTP vs FTP and
1126 similar.
1127
1128 22. Next major release
1129
1130 22.1 cleanup return codes
1131
1132 curl_easy_cleanup() returns void, but curl_multi_cleanup() returns a
1133 CURLMcode. These should be changed to be the same.
1134
1135 22.2 remove obsolete defines
1136
1137 remove obsolete defines from curl/curl.h
1138
1139 22.3 size_t
1140
1141 make several functions use size_t instead of int in their APIs
1142
1143 22.4 remove several functions
1144
1145 remove the following functions from the public API:
1146
1147 curl_getenv
1148
1149 curl_mprintf (and variations)
1150
1151 curl_strequal
1152
1153 curl_strnequal
1154
1155 They will instead become curlx_ - alternatives. That makes the curl app
1156 still capable of using them, by building with them from source.
1157
1158 These functions have no purpose anymore:
1159
1160 curl_multi_socket
1161
1162 curl_multi_socket_all
1163
1164 22.5 remove CURLOPT_FAILONERROR
1165
1166 Remove support for CURLOPT_FAILONERROR, it has gotten too kludgy and weird
1167 internally. Let the app judge success or not for itself.
1168
1169 22.7 remove progress meter from libcurl
1170
1171 The internally provided progress meter output doesn't belong in the library.
1172 Basically no application wants it (apart from curl) but instead applications
1173 can and should do their own progress meters using the progress callback.
1174
1175 The progress callback should then be bumped as well to get proper 64bit
1176 variable types passed to it instead of doubles so that big files work
1177 correctly.
1178
1179 22.8 remove 'curl_httppost' from public
1180
1181 curl_formadd() was made to fill in a public struct, but the fact that the
1182 struct is public is never really used by application for their own advantage
1183 but instead often restricts how the form functions can or can't be modified.
1184
1185 Changing them to return a private handle will benefit the implementation and
1186 allow us much greater freedoms while still maintaining a solid API and ABI.