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 All bugs documented in the KNOWN_BUGS document are subject for fixing!
13
14 1. libcurl
15 1.2 More data sharing
16 1.3 struct lifreq
17 1.4 signal-based resolver timeouts
18 1.5 get rid of PATH_MAX
19 1.6 progress callback without doubles
20
21 2. libcurl - multi interface
22 2.1 More non-blocking
23 2.2 Remove easy interface internally
24 2.4 Fix HTTP Pipelining for PUT
25
26 3. Documentation
27 3.1  More and better
28
29 4. FTP
30 4.1 HOST
31 4.2 Alter passive/active on failure and retry
32 4.3 Earlier bad letter detection
33 4.4 REST for large files
34 4.5 FTP proxy support
35 4.6 ASCII support
36
37 5. HTTP
38 5.1 Better persistency for HTTP 1.0
39 5.2 support FF3 sqlite cookie files
40 5.3 Rearrange request header order
41
42 6. TELNET
43 6.1 ditch stdin
44 6.2 ditch telnet-specific select
45 6.3 feature negotiation debug data
46 6.4 send data in chunks
47
48 7. SSL
49 7.1 Disable specific versions
50 7.2 Provide mutex locking API
51 7.3 Evaluate SSL patches
52 7.4 Cache OpenSSL contexts
53 7.5 Export session ids
54 7.6 Provide callback for cert verification
55 7.7 Support other SSL libraries
56 7.9 improve configure --with-ssl
57
58 8. GnuTLS
59 8.1 SSL engine stuff
60 8.3 check connection
61 8.4 non-gcrypt
62
63 9. Other protocols
64
65 10. New protocols
66 10.1 RSYNC
67
68 11. Client
69 11.1 sync
70 11.2 glob posts
71 11.3 prevent file overwriting
72 11.4 simultaneous parallel transfers
73 11.5 provide formpost headers
74 11.6 url-specific options
75 11.7 metalink support
76 11.8 warning when setting an option
77 11.9 IPv6 addresses with globbing
78
79 12. Build
80 12.1 roffit
81
82 13. Test suite
83 13.1 SSL tunnel
84 13.2 nicer lacking perl message
85 13.3 more protocols supported
86 13.4 more platforms supported
87
88 14. Next SONAME bump
89 14.1 http-style HEAD output for ftp
90 14.2 combine error codes
91 14.3 extend CURLOPT_SOCKOPTFUNCTION prototype
92
93 15. Next major release
94 15.1 cleanup return codes
95 15.2 remove obsolete defines
96 15.3 size_t
97 15.4 remove several functions
98 15.5 remove CURLOPT_FAILONERROR
99 15.6 remove CURLOPT_DNS_USE_GLOBAL_CACHE
100 15.7 remove progress meter from libcurl
101 15.8 remove 'curl_httppost' from public
102 15.9 have form functions use CURL handle argument
103
104==============================================================================
105
1061. libcurl
107
1081.2 More data sharing
109
110 curl_share_* functions already exist and work, and they can be extended to
111 share more. For example, enable sharing of the ares channel and the
112 connection cache.
113
1141.3 struct lifreq
115
116 Use 'struct lifreq' and SIOCGLIFADDR instead of 'struct ifreq' and
117 SIOCGIFADDR on newer Solaris versions as they claim the latter is obsolete.
118 To support ipv6 interface addresses for network interfaces properly.
119
1201.4 signal-based resolver timeouts
121
122 libcurl built without an asynchronous resolver library uses alarm() to time
123 out DNS lookups. When a timeout occurs, this causes libcurl to jump from the
124 signal handler back into the library with a sigsetjmp, which effectively
125 causes libcurl to continue running within the signal handler. This is
126 non-portable and could cause problems on some platforms. A discussion on the
127 problem is available at http://curl.haxx.se/mail/lib-2008-09/0197.html
128
129 Also, alarm() provides timeout resolution only to the nearest second. alarm
130 ought to be replaced by setitimer on systems that support it.
131
1321.5 get rid of PATH_MAX
133
134 Having code use and rely on PATH_MAX is not nice:
135 http://insanecoding.blogspot.com/2007/11/pathmax-simply-isnt.html
136
137 Currently the SSH based code uses it a bit, but to remove PATH_MAX from there
138 we need libssh2 to properly tell us when we pass in a too small buffer and
139 its current API (as of libssh2 1.2.7) doesn't.
140
1411.6 progress callback without doubles
142
143 The progress callback was introduced way back in the days and the choice to
144 use doubles in the arguments was possibly good at the time. Today the doubles
145 only confuse users and make the amounts less precise. We should introduce
146 another progress callback option that take precedence over the old one and
147 have both co-exist for a forseeable time until we can remove the double-using
148 one.
149
1502. libcurl - multi interface
151
1522.1 More non-blocking
153
154 Make sure we don't ever loop because of non-blocking sockets returning
155 EWOULDBLOCK or similar. Blocking cases include:
156
157 - Name resolves on non-windows unless c-ares is used
158 - NSS SSL connections
159 - Active FTP connections
160 - HTTP proxy CONNECT operations
161 - SOCKS proxy handshakes
162 - file:// transfers
163 - TELNET transfers
164 - The "DONE" operation (post transfer protocol-specific actions) for the
165   protocols SFTP, SMTP, FTP. Fixing Curl_done() for this is a worthy task.
166
1672.2 Remove easy interface internally
168
169 Make curl_easy_perform() a wrapper-function that simply creates a multi
170 handle, adds the easy handle to it, runs curl_multi_perform() until the
171 transfer is done, then detach the easy handle, destroy the multi handle and
172 return the easy handle's return code. This will thus make everything
173 internally use and assume the multi interface. The select()-loop should use
174 curl_multi_socket().
175
1762.4 Fix HTTP Pipelining for PUT
177
178 HTTP Pipelining can be a way to greatly enhance performance for multiple
179 serial requests and currently libcurl only supports that for HEAD and GET
180 requests but it should also be possible for PUT.
181
1823. Documentation
183
1843.1  More and better
185
186 Exactly
187
1884. FTP
189
1904.1 HOST
191
192 HOST is a suggested command in the works for a client to tell which host name
193 to use, to offer FTP servers named-based virtual hosting:
194
195 http://tools.ietf.org/html/draft-hethmon-mcmurray-ftp-hosts-11
196
1974.2 Alter passive/active on failure and retry
198
199 When trying to connect passively to a server which only supports active
200 connections, libcurl returns CURLE_FTP_WEIRD_PASV_REPLY and closes the
201 connection. There could be a way to fallback to an active connection (and
202 vice versa). http://curl.haxx.se/bug/feature.cgi?id=1754793
203
2044.3 Earlier bad letter detection
205
206 Make the detection of (bad) %0d and %0a codes in FTP url parts earlier in the
207 process to avoid doing a resolve and connect in vain.
208
2094.4 REST for large files
210
211 REST fix for servers not behaving well on >2GB requests. This should fail if
212 the server doesn't set the pointer to the requested index. The tricky
213 (impossible?) part is to figure out if the server did the right thing or not.
214
2154.5 FTP proxy support
216
217 Support the most common FTP proxies, Philip Newton provided a list allegedly
218 from ncftp. This is not a subject without debate, and is probably not really
219 suitable for libcurl.  http://curl.haxx.se/mail/archive-2003-04/0126.html
220
2214.6 ASCII support
222
223 FTP ASCII transfers do not follow RFC959. They don't convert the data
224 accordingly.
225
2265. HTTP
227
2285.1 Better persistency for HTTP 1.0
229
230 "Better" support for persistent connections over HTTP 1.0
231 http://curl.haxx.se/bug/feature.cgi?id=1089001
232
2335.2 support FF3 sqlite cookie files
234
235 Firefox 3 is changing from its former format to a a sqlite database instead.
236 We should consider how (lib)curl can/should support this.
237 http://curl.haxx.se/bug/feature.cgi?id=1871388
238
2395.3 Rearrange request header order
240
241 Server implementors often make an effort to detect browser and to reject
242 clients it can detect to not match. One of the last details we cannot yet
243 control in libcurl's HTTP requests, which also can be exploited to detect
244 that libcurl is in fact used even when it tries to impersonate a browser, is
245 the order of the request headers. I propose that we introduce a new option in
246 which you give headers a value, and then when the HTTP request is built it
247 sorts the headers based on that number. We could then have internally created
248 headers use a default value so only headers that need to be moved have to be
249 specified.
250
251
2526. TELNET
253
2546.1 ditch stdin
255
256Reading input (to send to the remote server) on stdin is a crappy solution for
257library purposes. We need to invent a good way for the application to be able
258to provide the data to send.
259
2606.2 ditch telnet-specific select
261
262 Move the telnet support's network select() loop go away and merge the code
263 into the main transfer loop. Until this is done, the multi interface won't
264 work for telnet.
265
2666.3 feature negotiation debug data
267
268  Add telnet feature negotiation data to the debug callback as header data.
269
2706.4 send data in chunks
271
272  Currently, telnet sends data one byte at a time.  This is fine for interactive
273  use, but inefficient for any other.  Sent data should be sent in larger
274  chunks.
275
2767. SSL
277
2787.1 Disable specific versions
279
280 Provide an option that allows for disabling specific SSL versions, such as
281 SSLv2 http://curl.haxx.se/bug/feature.cgi?id=1767276
282
2837.2 Provide mutex locking API
284
285 Provide a libcurl API for setting mutex callbacks in the underlying SSL
286 library, so that the same application code can use mutex-locking
287 independently of OpenSSL or GnutTLS being used.
288
2897.3 Evaluate SSL patches
290
291 Evaluate/apply Gertjan van Wingerde's SSL patches:
292 http://curl.haxx.se/mail/lib-2004-03/0087.html
293
2947.4 Cache OpenSSL contexts
295
296 "Look at SSL cafile - quick traces look to me like these are done on every
297 request as well, when they should only be necessary once per ssl context (or
298 once per handle)". The major improvement we can rather easily do is to make
299 sure we don't create and kill a new SSL "context" for every request, but
300 instead make one for every connection and re-use that SSL context in the same
301 style connections are re-used. It will make us use slightly more memory but
302 it will libcurl do less creations and deletions of SSL contexts.
303
3047.5 Export session ids
305
306 Add an interface to libcurl that enables "session IDs" to get
307 exported/imported. Cris Bailiff said: "OpenSSL has functions which can
308 serialise the current SSL state to a buffer of your choice, and recover/reset
309 the state from such a buffer at a later date - this is used by mod_ssl for
310 apache to implement and SSL session ID cache".
311
3127.6 Provide callback for cert verification
313
314 OpenSSL supports a callback for customised verification of the peer
315 certificate, but this doesn't seem to be exposed in the libcurl APIs. Could
316 it be? There's so much that could be done if it were!
317
3187.7 Support other SSL libraries
319
320 Make curl's SSL layer capable of using other free SSL libraries.  Such as
321 MatrixSSL (http://www.matrixssl.org/).
322
3237.9 improve configure --with-ssl
324
325 make the configure --with-ssl option first check for OpenSSL, then GnuTLS,
326 then NSS...
327
3288. GnuTLS
329
3308.1 SSL engine stuff
331
332 Is this even possible?
333
3348.3 check connection
335
336 Add a way to check if the connection seems to be alive, to correspond to the
337 SSL_peak() way we use with OpenSSL.
338
3398.4 non-gcrypt
340
341 libcurl assumes that there are gcrypt functions available when
342 GnuTLS is.
343
344 GnuTLS can be built to use libnettle instead as crypto library,
345 which breaks the previously mentioned assumption
346
347 The correct fix would be to detect which crypto layer that is in use and
348 adapt our code to use that instead of blindly assuming gcrypt.
349
3509. Other protocols
351
35210. New protocols
353
35410.1 RSYNC
355
356 There's no RFC for protocol nor URI/URL format.  An implementation should
357 most probably use an existing rsync library, such as librsync.
358
35911. Client
360
36111.1 sync
362
363 "curl --sync http://example.com/feed[1-100].rss" or
364 "curl --sync http://example.net/{index,calendar,history}.html"
365
366 Downloads a range or set of URLs using the remote name, but only if the
367 remote file is newer than the local file. A Last-Modified HTTP date header
368 should also be used to set the mod date on the downloaded file.
369
37011.2 glob posts
371
372 Globbing support for -d and -F, as in 'curl -d "name=foo[0-9]" URL'.
373 This is easily scripted though.
374
37511.3 prevent file overwriting
376
377 Add an option that prevents cURL from overwriting existing local files. When
378 used, and there already is an existing file with the target file name
379 (either -O or -o), a number should be appended (and increased if already
380 existing). So that index.html becomes first index.html.1 and then
381 index.html.2 etc.
382
38311.4 simultaneous parallel transfers
384
385 The client could be told to use maximum N simultaneous parallel transfers and
386 then just make sure that happens. It should of course not make more than one
387 connection to the same remote host. This would require the client to use the
388 multi interface. http://curl.haxx.se/bug/feature.cgi?id=1558595
389
39011.5 provide formpost headers
391
392 Extending the capabilities of the multipart formposting. How about leaving
393 the ';type=foo' syntax as it is and adding an extra tag (headers) which
394 works like this: curl -F "coolfiles=@fil1.txt;headers=@fil1.hdr" where
395 fil1.hdr contains extra headers like
396
397   Content-Type: text/plain; charset=KOI8-R"
398   Content-Transfer-Encoding: base64
399   X-User-Comment: Please don't use browser specific HTML code
400
401 which should overwrite the program reasonable defaults (plain/text,
402 8bit...)
403
40411.6 url-specific options
405
406 Provide a way to make options bound to a specific URL among several on the
407 command line. Possibly by letting ':' separate options between URLs,
408 similar to this:
409
410    curl --data foo --url url.com : \
411        --url url2.com : \
412        --url url3.com --data foo3
413
414 (More details: http://curl.haxx.se/mail/archive-2004-07/0133.html)
415
416 The example would do a POST-GET-POST combination on a single command line.
417
41811.7 metalink support
419
420 Add metalink support to curl (http://www.metalinker.org/). This is most useful
421 with simultaneous parallel transfers (11.6) but not necessary.
422
42311.8 warning when setting an option
424
425  Display a warning when libcurl returns an error when setting an option.
426  This can be useful to tell when support for a particular feature hasn't been
427  compiled into the library.
428
42911.9 IPv6 addresses with globbing
430
431  Currently the command line client needs to get url globbing disabled (with
432  -g) for it to support IPv6 numerical addresses. This is a rather silly flaw
433  that should be corrected. It probably involves a smarter detection of the
434  '[' and ']' letters.
435
43612. Build
437
43812.1 roffit
439
440 Consider extending 'roffit' to produce decent ASCII output, and use that
441 instead of (g)nroff when building src/hugehelp.c
442
44313. Test suite
444
44513.1 SSL tunnel
446
447 Make our own version of stunnel for simple port forwarding to enable HTTPS
448 and FTP-SSL tests without the stunnel dependency, and it could allow us to
449 provide test tools built with either OpenSSL or GnuTLS
450
45113.2 nicer lacking perl message
452
453 If perl wasn't found by the configure script, don't attempt to run the tests
454 but explain something nice why it doesn't.
455
45613.3 more protocols supported
457
458 Extend the test suite to include more protocols. The telnet could just do ftp
459 or http operations (for which we have test servers).
460
46113.4 more platforms supported
462
463 Make the test suite work on more platforms. OpenBSD and Mac OS. Remove
464 fork()s and it should become even more portable.
465
46614. Next SONAME bump
467
46814.1 http-style HEAD output for ftp
469
470 #undef CURL_FTP_HTTPSTYLE_HEAD in lib/ftp.c to remove the HTTP-style headers
471 from being output in NOBODY requests over ftp
472
47314.2 combine error codes
474
475 Combine some of the error codes to remove duplicates.  The original
476 numbering should not be changed, and the old identifiers would be
477 macroed to the new ones in an CURL_NO_OLDIES section to help with
478 backward compatibility.
479
480 Candidates for removal and their replacements:
481
482    CURLE_FILE_COULDNT_READ_FILE => CURLE_REMOTE_FILE_NOT_FOUND
483    CURLE_FTP_COULDNT_RETR_FILE => CURLE_REMOTE_FILE_NOT_FOUND
484    CURLE_FTP_COULDNT_USE_REST => CURLE_RANGE_ERROR
485    CURLE_FUNCTION_NOT_FOUND => CURLE_FAILED_INIT
486    CURLE_LDAP_INVALID_URL => CURLE_URL_MALFORMAT
487    CURLE_TFTP_NOSUCHUSER => CURLE_TFTP_ILLEGAL
488    CURLE_TFTP_NOTFOUND => CURLE_REMOTE_FILE_NOT_FOUND
489    CURLE_TFTP_PERM => CURLE_REMOTE_ACCESS_DENIED
490
49114.3 extend CURLOPT_SOCKOPTFUNCTION prototype
492
493 The current prototype only provides 'purpose' that tells what the
494 connection/socket is for, but not any protocol or similar. It makes it hard
495 for applications to differentiate on TCP vs UDP and even HTTP vs FTP and
496 similar.
497
49815. Next major release
499
50015.1 cleanup return codes
501
502 curl_easy_cleanup() returns void, but curl_multi_cleanup() returns a
503 CURLMcode. These should be changed to be the same.
504
50515.2 remove obsolete defines
506
507 remove obsolete defines from curl/curl.h
508
50915.3 size_t
510
511 make several functions use size_t instead of int in their APIs
512
51315.4 remove several functions
514
515 remove the following functions from the public API:
516
517 curl_getenv
518
519 curl_mprintf (and variations)
520
521 curl_strequal
522
523 curl_strnequal
524
525 They will instead become curlx_ - alternatives. That makes the curl app
526 still capable of using them, by building with them from source.
527
528 These functions have no purpose anymore:
529
530 curl_multi_socket
531
532 curl_multi_socket_all
533
53415.5 remove CURLOPT_FAILONERROR
535
536 Remove support for CURLOPT_FAILONERROR, it has gotten too kludgy and weird
537 internally. Let the app judge success or not for itself.
538
53915.6 remove CURLOPT_DNS_USE_GLOBAL_CACHE
540
541 Remove support for a global DNS cache. Anything global is silly, and we
542 already offer the share interface for the same functionality but done
543 "right".
544
54515.7 remove progress meter from libcurl
546
547 The internally provided progress meter output doesn't belong in the library.
548 Basically no application wants it (apart from curl) but instead applications
549 can and should do their own progress meters using the progress callback.
550
551 The progress callback should then be bumped as well to get proper 64bit
552 variable types passed to it instead of doubles so that big files work
553 correctly.
554
55515.8 remove 'curl_httppost' from public
556
557 curl_formadd() was made to fill in a public struct, but the fact that the
558 struct is public is never really used by application for their own advantage
559 but instead often restricts how the form functions can or can't be modified.
560
561 Changing them to return a private handle will benefit the implementation and
562 allow us much greater freedoms while still maintining a solid API and ABI.
563
56415.9 have form functions use CURL handle argument
565
566 curl_formadd() and curl_formget() both currently have no CURL handle
567 argument, but both can use a callback that is set in the easy handle, and
568 thus curl_formget() with callback cannot function without first having
569 curl_easy_perform() (or similar) called - which is hard to grasp and a design
570 mistake.
571