aboutsummaryrefslogtreecommitdiff
path: root/docs/TODO
blob: cf1a769c317beca2223e9628c80dc5bf6462d87e (plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
1001
1002
1003
1004
1005
1006
1007
1008
1009
1010
1011
1012
1013
1014
1015
1016
1017
1018
1019
1020
1021
1022
1023
1024
1025
1026
1027
1028
1029
1030
1031
1032
1033
1034
1035
1036
1037
1038
1039
1040
1041
1042
1043
1044
1045
1046
1047
1048
1049
1050
1051
1052
1053
1054
1055
1056
1057
1058
1059
1060
1061
1062
1063
1064
1065
1066
1067
1068
1069
1070
1071
1072
1073
1074
1075
1076
1077
1078
1079
1080
1081
1082
1083
1084
1085
1086
1087
1088
1089
1090
1091
1092
1093
1094
1095
1096
1097
1098
1099
1100
1101
1102
1103
1104
1105
1106
1107
1108
1109
1110
1111
1112
1113
1114
1115
1116
1117
1118
1119
1120
1121
1122
1123
1124
1125
1126
1127
1128
1129
1130
1131
1132
1133
1134
1135
1136
1137
1138
1139
1140
1141
1142
1143
1144
1145
1146
1147
1148
1149
1150
1151
1152
1153
1154
1155
1156
1157
1158
1159
1160
1161
1162
1163
1164
1165
1166
1167
1168
1169
1170
1171
1172
1173
1174
1175
1176
1177
1178
1179
1180
1181
1182
1183
1184
1185
1186
1187
1188
1189
1190
1191
1192
1193
1194
1195
1196
1197
1198
1199
1200
1201
1202
1203
1204
1205
1206
1207
1208
1209
1210
1211
1212
1213
1214
1215
1216
1217
1218
1219
1220
1221
1222
1223
1224
1225
1226
1227
1228
1229
1230
1231
1232
1233
1234
1235
1236
1237
1238
1239
1240
1241
1242
1243
1244
1245
1246
1247
1248
1249
1250
1251
1252
1253
1254
1255
1256
1257
1258
1259
1260
1261
1262
1263
1264
1265
1266
1267
1268
1269
1270
1271
1272
1273
1274
1275
1276
1277
1278
                                  _   _ ____  _
                              ___| | | |  _ \| |
                             / __| | | | |_) | |
                            | (__| |_| |  _ <| |___
                             \___|\___/|_| \_\_____|

                Things that could be nice to do in the future

 Things to do in project curl. Please tell us what you think, contribute and
 send us patches that improve things!

 Be aware that these are things that we could do, or have once been considered
 things we could do. If you want to work on any of these areas, please
 consider bringing it up for discussions first on the mailing list so that we
 all agree it is still a good idea for the project!

 All bugs documented in the KNOWN_BUGS document are subject for fixing!

 1. libcurl
 1.2 More data sharing
 1.3 struct lifreq
 1.4 signal-based resolver timeouts
 1.5 get rid of PATH_MAX
 1.6 Modified buffer size approach
 1.7 Support HTTP/2 for HTTP(S) proxies
 1.8 CURLOPT_RESOLVE for any port number
 1.9 Cache negative name resolves
 1.10 auto-detect proxy
 1.11 minimize dependencies with dynamically loaded modules
 1.12 updated DNS server while running
 1.13 DNS-over-HTTPS
 1.14 Typesafe curl_easy_setopt()
 1.15 Monitor connections in the connection pool
 1.16 Try to URL encode given URL
 1.17 Add support for IRIs
 1.18 try next proxy if one doesn't work
 1.19 Timeout idle connections from the pool
 1.20 SRV and URI DNS records
 1.21 API for URL parsing/splitting
 1.22 CURLINFO_PAUSE_STATE
 1.23 Offer API to flush the connection pool
 1.24 TCP Fast Open for windows
 1.25 Expose tried IP addresses that failed
 1.26 CURL_REFUSE_CLEARTEXT
 1.27 hardcode the "localhost" addresses
 1.28 FD_CLOEXEC

 2. libcurl - multi interface
 2.1 More non-blocking
 2.2 Better support for same name resolves
 2.3 Non-blocking curl_multi_remove_handle()
 2.4 Split connect and authentication process
 2.5 Edge-triggered sockets should work

 3. Documentation
 3.2 Provide cmake config-file

 4. FTP
 4.1 HOST
 4.2 Alter passive/active on failure and retry
 4.3 Earlier bad letter detection
 4.4 REST for large files
 4.5 ASCII support
 4.6 GSSAPI via Windows SSPI
 4.7 STAT for LIST without data connection
 4.8 Option to ignore private IP addresses in PASV response

 5. HTTP
 5.1 Better persistency for HTTP 1.0
 5.2 support FF3 sqlite cookie files
 5.3 Rearrange request header order
 5.5 auth= in URLs
 5.6 Refuse "downgrade" redirects
 5.7 QUIC
 5.8 Leave secure cookies alone

 6. TELNET
 6.1 ditch stdin
 6.2 ditch telnet-specific select
 6.3 feature negotiation debug data

 7. SMTP
 7.1 Pipelining
 7.2 Enhanced capability support
 7.3 Add CURLOPT_MAIL_CLIENT option

 8. POP3
 8.1 Pipelining
 8.2 Enhanced capability support

 9. IMAP
 9.1 Enhanced capability support

 10. LDAP
 10.1 SASL based authentication mechanisms

 11. SMB
 11.1 File listing support
 11.2 Honor file timestamps
 11.3 Use NTLMv2
 11.4 Create remote directories

 12. New protocols
 12.1 RSYNC

 13. SSL
 13.1 Disable specific versions
 13.2 Provide mutex locking API
 13.3 Support in-memory certs/ca certs/keys
 13.4 Cache/share OpenSSL contexts
 13.5 Export session ids
 13.6 Provide callback for cert verification
 13.7 improve configure --with-ssl
 13.8 Support DANE
 13.9 Configurable loading of OpenSSL configuration file
 13.10 Support Authority Information Access certificate extension (AIA)
 13.11 Support intermediate & root pinning for PINNEDPUBLICKEY
 13.12 Support HSTS
 13.13 Support HPKP
 13.14 Support the clienthello extension

 14. GnuTLS
 14.1 SSL engine stuff
 14.2 check connection

 15. WinSSL/SChannel
 15.1 Add support for client certificate authentication
 15.3 Add support for the --ciphers option

 16. SASL
 16.1 Other authentication mechanisms
 16.2 Add QOP support to GSSAPI authentication
 16.3 Support binary messages (i.e.: non-base64)

 17. SSH protocols
 17.1 Multiplexing
 17.2 SFTP performance
 17.3 Support better than MD5 hostkey hash
 17.4 Support CURLOPT_PREQUOTE

 18. Command line tool
 18.1 sync
 18.2 glob posts
 18.3 prevent file overwriting
 18.4 simultaneous parallel transfers
 18.5 UTF-8 filenames in Content-Disposition
 18.6 warning when setting an option
 18.7 warning if curl version is not in sync with libcurl version
 18.8 offer color-coded HTTP header output
 18.9 Choose the name of file in braces for complex URLs
 18.10 improve how curl works in a windows console window
 18.11 -w output to stderr
 18.12 keep running, read instructions from pipe/socket
 18.13 support metalink in http headers
 18.14 --fail without --location should treat 3xx as a failure
 18.15 --retry should resume
 18.16 send only part of --data
 18.17 consider file name from the redirected URL with -O ?
 18.18 retry on network is unreachable
 18.19 expand ~/ in config files
 18.20 host name sections in config files

 19. Build
 19.1 roffit
 19.2 Enable PIE and RELRO by default

 20. Test suite
 20.1 SSL tunnel
 20.2 nicer lacking perl message
 20.3 more protocols supported
 20.4 more platforms supported
 20.5 Add support for concurrent connections
 20.6 Use the RFC6265 test suite

 21. Next SONAME bump
 21.1 http-style HEAD output for FTP
 21.2 combine error codes
 21.3 extend CURLOPT_SOCKOPTFUNCTION prototype

 22. Next major release
 22.1 cleanup return codes
 22.2 remove obsolete defines
 22.3 size_t
 22.4 remove several functions
 22.5 remove CURLOPT_FAILONERROR
 22.6 remove CURLOPT_DNS_USE_GLOBAL_CACHE
 22.7 remove progress meter from libcurl
 22.8 remove 'curl_httppost' from public

==============================================================================

1. libcurl

1.2 More data sharing

 curl_share_* functions already exist and work, and they can be extended to
 share more. For example, enable sharing of the ares channel.

1.3 struct lifreq

 Use 'struct lifreq' and SIOCGLIFADDR instead of 'struct ifreq' and
 SIOCGIFADDR on newer Solaris versions as they claim the latter is obsolete.
 To support IPv6 interface addresses for network interfaces properly.

1.4 signal-based resolver timeouts

 libcurl built without an asynchronous resolver library uses alarm() to time
 out DNS lookups. When a timeout occurs, this causes libcurl to jump from the
 signal handler back into the library with a sigsetjmp, which effectively
 causes libcurl to continue running within the signal handler. This is
 non-portable and could cause problems on some platforms. A discussion on the
 problem is available at https://curl.haxx.se/mail/lib-2008-09/0197.html

 Also, alarm() provides timeout resolution only to the nearest second. alarm
 ought to be replaced by setitimer on systems that support it.

1.5 get rid of PATH_MAX

 Having code use and rely on PATH_MAX is not nice:
 https://insanecoding.blogspot.com/2007/11/pathmax-simply-isnt.html

 Currently the SSH based code uses it a bit, but to remove PATH_MAX from there
 we need libssh2 to properly tell us when we pass in a too small buffer and
 its current API (as of libssh2 1.2.7) doesn't.

1.6 Modified buffer size approach

 Current libcurl allocates a fixed 16K size buffer for download and an
 additional 16K for upload. They are always unconditionally part of the easy
 handle. If CRLF translations are requested, an additional 32K "scratch
 buffer" is allocated. A total of 64K transfer buffers in the worst case.

 First, while the handles are not actually in use these buffers could be freed
 so that lingering handles just kept in queues or whatever waste less memory.

 Secondly, SFTP is a protocol that needs to handle many ~30K blocks at once
 since each need to be individually acked and therefore libssh2 must be
 allowed to send (or receive) many separate ones in parallel to achieve high
 transfer speeds. A current libcurl build with a 16K buffer makes that
 impossible, but one with a 512K buffer will reach MUCH faster transfers. But
 allocating 512K unconditionally for all buffers just in case they would like
 to do fast SFTP transfers at some point is not a good solution either.

 Dynamically allocate buffer size depending on protocol in use in combination
 with freeing it after each individual transfer? Other suggestions?

1.7 Support HTTP/2 for HTTP(S) proxies

 Support for doing HTTP/2 to HTTP and HTTPS proxies is still missing.

1.8 CURLOPT_RESOLVE for any port number

 This option allows applications to set a replacement IP address for a given
 host + port pair. Consider making support for providing a replacement address
 for the host name on all port numbers.

 See https://github.com/curl/curl/issues/1264

1.9 Cache negative name resolves

 A name resolve that has failed is likely to fail when made again within a
 short period of time. Currently we only cache positive responses.

1.10 auto-detect proxy

 libcurl could be made to detect the system proxy setup automatically and use
 that. On Windows, macOS and Linux desktops for example.

 The pull-request to use libproxy for this was deferred due to doubts on the
 reliability of the dependency and how to use it:
 https://github.com/curl/curl/pull/977

 libdetectproxy is a (C++) library for detecting the proxy on Windows
 https://github.com/paulharris/libdetectproxy

1.11 minimize dependencies with dynamically loaded modules

 We can create a system with loadable modules/plug-ins, where these modules
 would be the ones that link to 3rd party libs. That would allow us to avoid
 having to load ALL dependencies since only the necessary ones for this
 app/invoke/used protocols would be necessary to load.  See
 https://github.com/curl/curl/issues/349

1.12 updated DNS server while running

 If /etc/resolv.conf gets updated while a program using libcurl is running, it
 is may cause name resolves to fail unless res_init() is called. We should
 consider calling res_init() + retry once unconditionally on all name resolve
 failures to mitigate against this. Firefox works like that. Note that Windows
 doesn't have res_init() or an alternative.

 https://github.com/curl/curl/issues/2251

1.13 DNS-over-HTTPS

 By adding support for DNS-over-HTTPS curl could resolve host names using a
 totally separate name server than the standard system resolver, while at the
 same time doing so over a communication channel that enhances privacy and
 security.

 https://github.com/curl/curl/wiki/DNS-over-HTTPS

1.14 Typesafe curl_easy_setopt()

 One of the most common problems in libcurl using applications is the lack of
 type checks for curl_easy_setopt() which happens because it accepts varargs
 and thus can take any type.

 One possible solution to this is to introduce a few different versions of the
 setopt version for the different kinds of data you can set.

  curl_easy_set_num() - sets a long value

  curl_easy_set_large() - sets a curl_off_t value

  curl_easy_set_ptr() - sets a pointer

  curl_easy_set_cb() - sets a callback PLUS its callback data

1.15 Monitor connections in the connection pool

 libcurl's connection cache or pool holds a number of open connections for the
 purpose of possible subsequent connection reuse. It may contain a few up to a
 significant amount of connections. Currently, libcurl leaves all connections
 as they are and first when a connection is iterated over for matching or
 reuse purpose it is verified that it is still alive.

 Those connections may get closed by the server side for idleness or they may
 get a HTTP/2 ping from the peer to verify that they're still alive. By adding
 monitoring of the connections while in the pool, libcurl can detect dead
 connections (and close them) better and earlier, and it can handle HTTP/2
 pings to keep such ones alive even when not actively doing transfers on them.

1.16 Try to URL encode given URL

 Given a URL that for example contains spaces, libcurl could have an option
 that would try somewhat harder than it does now and convert spaces to %20 and
 perhaps URL encoded byte values over 128 etc (basically do what the redirect
 following code already does).

 https://github.com/curl/curl/issues/514

1.17 Add support for IRIs

 IRIs (RFC 3987) allow localized, non-ascii, names in the URL. To properly
 support this, curl/libcurl would need to translate/encode the given input
 from the input string encoding into percent encoded output "over the wire".

 To make that work smoothly for curl users even on Windows, curl would
 probably need to be able to convert from several input encodings.

1.18 try next proxy if one doesn't work

 Allow an application to specify a list of proxies to try, and failing to
 connect to the first go on and try the next instead until the list is
 exhausted. Browsers support this feature at least when they specify proxies
 using PACs.

 https://github.com/curl/curl/issues/896

1.19 Timeout idle connections from the pool

 libcurl currently keeps connections in its connection pool for an indefinite
 period of time, until it either gets reused, gets noticed that it has been
 closed by the server or gets pruned to make room for a new connection.

 To reduce overhead (especially for when we add monitoring of the connections
 in the pool), we should introduce a timeout so that connections that have
 been idle for N seconds get closed.

1.20 SRV and URI DNS records

 Offer support for resolving SRV and URI DNS records for libcurl to know which
 server to connect to for various protocols (including HTTP!).

1.21 API for URL parsing/splitting

 libcurl has always parsed URLs internally and never exposed any API or
 features to allow applications to do it. Still most or many applications
 using libcurl need that ability. In polls to users, we've learned that many
 libcurl users would like to see and use such an API.

1.22 CURLINFO_PAUSE_STATE

 Return information about the transfer's current pause state, in both
 directions. https://github.com/curl/curl/issues/2588

1.23 Offer API to flush the connection pool

 Sometimes applications want to flush all the existing connections kept alive.
 An API could allow a forced flush or just a forced loop that would properly
 close all connections that have been closed by the server already.

1.24 TCP Fast Open for windows

 libcurl supports the CURLOPT_TCP_FASTOPEN option since 7.49.0 for Linux and
 Mac OS. Windows supports TCP Fast Open starting with Windows 10, version 1607
 and we should add support for it.

1.25 Expose tried IP addresses that failed

 When libcurl fails to connect to a host, it should be able to offer the
 application the list of IP addresses that were used in the attempt.

 https://github.com/curl/curl/issues/2126

1.26 CURL_REFUSE_CLEARTEXT

 An environment variable that when set will make libcurl refuse to use any
 cleartext network protocol. That's all non-encrypted ones (FTP, HTTP, Gopher,
 etc). By adding the check to libcurl and not just curl, this environment
 variable can then help users to block all libcurl-using programs from
 accessing the network using unsafe protocols.

 The variable could be given some sort of syntax or different levels and be
 used to also allow for example users to refuse libcurl to do transfers with
 HTTPS certificate checks disabled.

 It could also offer to refuse usernames in URLs (see TODO 1.1)

1.27 hardcode the "localhost" addresses

 There's this new spec getting adopted that says "localhost" should always and
 unconditionally be a local address and not get resolved by a DNS server. A
 fine way for curl to fix this would be to simply hard-code the response to
 127.0.0.1 and/or ::1 (depending on what IP versions that are requested). This
 is what the browsers probably will do with this hostname.

 https://bugzilla.mozilla.org/show_bug.cgi?id=1220810

 https://tools.ietf.org/html/draft-ietf-dnsop-let-localhost-be-localhost-02

1.28 FD_CLOEXEC

 It sets the close-on-exec flag for the file descriptor, which causes the file
 descriptor to be automatically (and atomically) closed when any of the
 exec-family functions succeed. Should probably be set by default?

 https://github.com/curl/curl/issues/2252

2. libcurl - multi interface

2.1 More non-blocking

 Make sure we don't ever loop because of non-blocking sockets returning
 EWOULDBLOCK or similar. Blocking cases include:

 - Name resolves on non-windows unless c-ares or the threaded resolver is used
 - SOCKS proxy handshakes
 - file:// transfers
 - TELNET transfers
 - The "DONE" operation (post transfer protocol-specific actions) for the
   protocols SFTP, SMTP, FTP. Fixing Curl_done() for this is a worthy task.

2.2 Better support for same name resolves

 If a name resolve has been initiated for name NN and a second easy handle
 wants to resolve that name as well, make it wait for the first resolve to end
 up in the cache instead of doing a second separate resolve. This is
 especially needed when adding many simultaneous handles using the same host
 name when the DNS resolver can get flooded.

2.3 Non-blocking curl_multi_remove_handle()

 The multi interface has a few API calls that assume a blocking behavior, like
 add_handle() and remove_handle() which limits what we can do internally. The
 multi API need to be moved even more into a single function that "drives"
 everything in a non-blocking manner and signals when something is done. A
 remove or add would then only ask for the action to get started and then
 multi_perform() etc still be called until the add/remove is completed.

2.4 Split connect and authentication process

 The multi interface treats the authentication process as part of the connect
 phase. As such any failures during authentication won't trigger the relevant
 QUIT or LOGOFF for protocols such as IMAP, POP3 and SMTP.

2.5 Edge-triggered sockets should work

 The multi_socket API should work with edge-triggered socket events. One of
 the internal actions that need to be improved for this to work perfectly is
 the 'maxloops' handling in transfer.c:readwrite_data().

3. Documentation

3.2 Provide cmake config-file

 A config-file package is a set of files provided by us to allow applications
 to write cmake scripts to find and use libcurl easier. See
 https://github.com/curl/curl/issues/885

4. FTP

4.1 HOST

 HOST is a command for a client to tell which host name to use, to offer FTP
 servers named-based virtual hosting:

 https://tools.ietf.org/html/rfc7151

4.2 Alter passive/active on failure and retry

 When trying to connect passively to a server which only supports active
 connections, libcurl returns CURLE_FTP_WEIRD_PASV_REPLY and closes the
 connection. There could be a way to fallback to an active connection (and
 vice versa). https://curl.haxx.se/bug/feature.cgi?id=1754793

4.3 Earlier bad letter detection

 Make the detection of (bad) %0d and %0a codes in FTP URL parts earlier in the
 process to avoid doing a resolve and connect in vain.

4.4 REST for large files

 REST fix for servers not behaving well on >2GB requests. This should fail if
 the server doesn't set the pointer to the requested index. The tricky
 (impossible?) part is to figure out if the server did the right thing or not.

4.5 ASCII support

 FTP ASCII transfers do not follow RFC959. They don't convert the data
 accordingly.

4.6 GSSAPI via Windows SSPI

 In addition to currently supporting the SASL GSSAPI mechanism (Kerberos V5)
 via third-party GSS-API libraries, such as Heimdal or MIT Kerberos, also add
 support for GSSAPI authentication via Windows SSPI.

4.7 STAT for LIST without data connection

 Some FTP servers allow STAT for listing directories instead of using LIST,
 and the response is then sent over the control connection instead of as the
 otherwise usedw data connection: http://www.nsftools.com/tips/RawFTP.htm#STAT

 This is not detailed in any FTP specification.

4.8 Option to ignore private IP addresses in PASV response

 Some servers respond with and some other FTP client implementations can
 ignore private (RFC 1918 style) IP addresses when received in PASV responses.
 To consider for libcurl as well. See https://github.com/curl/curl/issues/1455

5. HTTP

5.1 Better persistency for HTTP 1.0

 "Better" support for persistent connections over HTTP 1.0
 https://curl.haxx.se/bug/feature.cgi?id=1089001

5.2 support FF3 sqlite cookie files

 Firefox 3 is changing from its former format to a a sqlite database instead.
 We should consider how (lib)curl can/should support this.
 https://curl.haxx.se/bug/feature.cgi?id=1871388

5.3 Rearrange request header order

 Server implementors often make an effort to detect browser and to reject
 clients it can detect to not match. One of the last details we cannot yet
 control in libcurl's HTTP requests, which also can be exploited to detect
 that libcurl is in fact used even when it tries to impersonate a browser, is
 the order of the request headers. I propose that we introduce a new option in
 which you give headers a value, and then when the HTTP request is built it
 sorts the headers based on that number. We could then have internally created
 headers use a default value so only headers that need to be moved have to be
 specified.

5.5 auth= in URLs

 Add the ability to specify the preferred authentication mechanism to use by
 using ;auth=<mech> in the login part of the URL.

 For example:

 http://test:pass;auth=NTLM@example.com would be equivalent to specifying --user
 test:pass;auth=NTLM or --user test:pass --ntlm from the command line.

 Additionally this should be implemented for proxy base URLs as well.

5.6 Refuse "downgrade" redirects

 See https://github.com/curl/curl/issues/226

 Consider a way to tell curl to refuse to "downgrade" protocol with a redirect
 and/or possibly a bit that refuses redirect to change protocol completely.

5.7 QUIC

 The standardization process of QUIC has been taken to the IETF and can be
 followed on the [IETF QUIC Mailing
 list](https://www.ietf.org/mailman/listinfo/quic). I'd like us to get on the
 bandwagon. Ideally, this would be done with a separate library/project to
 handle the binary/framing layer in a similar fashion to how HTTP/2 is
 implemented. This, to allow other projects to benefit from the work and to
 thus broaden the interest and chance of others to participate.

5.8 Leave secure cookies alone

 Non-secure origins (HTTP sites) should not be allowed to set or modify
 cookies with the 'secure' property:

 https://tools.ietf.org/html/draft-ietf-httpbis-cookie-alone-01


6. TELNET

6.1 ditch stdin

Reading input (to send to the remote server) on stdin is a crappy solution for
library purposes. We need to invent a good way for the application to be able
to provide the data to send.

6.2 ditch telnet-specific select

 Move the telnet support's network select() loop go away and merge the code
 into the main transfer loop. Until this is done, the multi interface won't
 work for telnet.

6.3 feature negotiation debug data

  Add telnet feature negotiation data to the debug callback as header data.


7. SMTP

7.1 Pipelining

 Add support for pipelining emails.

7.2 Enhanced capability support

 Add the ability, for an application that uses libcurl, to obtain the list of
 capabilities returned from the EHLO command.

7.3 Add CURLOPT_MAIL_CLIENT option

 Rather than use the URL to specify the mail client string to present in the
 HELO and EHLO commands, libcurl should support a new CURLOPT specifically for
 specifying this data as the URL is non-standard and to be honest a bit of a
 hack ;-)

 Please see the following thread for more information:
 https://curl.haxx.se/mail/lib-2012-05/0178.html


8. POP3

8.1 Pipelining

 Add support for pipelining commands.

8.2 Enhanced capability support

 Add the ability, for an application that uses libcurl, to obtain the list of
 capabilities returned from the CAPA command.

9. IMAP

9.1 Enhanced capability support

 Add the ability, for an application that uses libcurl, to obtain the list of
 capabilities returned from the CAPABILITY command.

10. LDAP

10.1 SASL based authentication mechanisms

 Currently the LDAP module only supports ldap_simple_bind_s() in order to bind
 to an LDAP server. However, this function sends username and password details
 using the simple authentication mechanism (as clear text). However, it should
 be possible to use ldap_bind_s() instead specifying the security context
 information ourselves.

11. SMB

11.1 File listing support

Add support for listing the contents of a SMB share. The output should probably
be the same as/similar to FTP.

11.2 Honor file timestamps

The timestamp of the transferred file should reflect that of the original file.

11.3 Use NTLMv2

Currently the SMB authentication uses NTLMv1.

11.4 Create remote directories

Support for creating remote directories when uploading a file to a directory
that doesn't exist on the server, just like --ftp-create-dirs.

12. New protocols

12.1 RSYNC

 There's no RFC for the protocol or an URI/URL format.  An implementation
 should most probably use an existing rsync library, such as librsync.

13. SSL

13.1 Disable specific versions

 Provide an option that allows for disabling specific SSL versions, such as
 SSLv2 https://curl.haxx.se/bug/feature.cgi?id=1767276

13.2 Provide mutex locking API

 Provide a libcurl API for setting mutex callbacks in the underlying SSL
 library, so that the same application code can use mutex-locking
 independently of OpenSSL or GnutTLS being used.

13.3 Support in-memory certs/ca certs/keys

 You can specify the private and public keys for SSH/SSL as file paths. Some
 programs want to avoid using files and instead just pass them as in-memory
 data blobs. There's probably a challenge to make this work across the
 plethory of different TLS and SSH backends that curl supports.
 https://github.com/curl/curl/issues/2310

13.4 Cache/share OpenSSL contexts

 "Look at SSL cafile - quick traces look to me like these are done on every
 request as well, when they should only be necessary once per SSL context (or
 once per handle)". The major improvement we can rather easily do is to make
 sure we don't create and kill a new SSL "context" for every request, but
 instead make one for every connection and re-use that SSL context in the same
 style connections are re-used. It will make us use slightly more memory but
 it will libcurl do less creations and deletions of SSL contexts.

 Technically, the "caching" is probably best implemented by getting added to
 the share interface so that easy handles who want to and can reuse the
 context specify that by sharing with the right properties set.

 https://github.com/curl/curl/issues/1110

13.5 Export session ids

 Add an interface to libcurl that enables "session IDs" to get
 exported/imported. Cris Bailiff said: "OpenSSL has functions which can
 serialise the current SSL state to a buffer of your choice, and recover/reset
 the state from such a buffer at a later date - this is used by mod_ssl for
 apache to implement and SSL session ID cache".

13.6 Provide callback for cert verification

 OpenSSL supports a callback for customised verification of the peer
 certificate, but this doesn't seem to be exposed in the libcurl APIs. Could
 it be? There's so much that could be done if it were!

13.7 improve configure --with-ssl

 make the configure --with-ssl option first check for OpenSSL, then GnuTLS,
 then NSS...

13.8 Support DANE

 DNS-Based Authentication of Named Entities (DANE) is a way to provide SSL
 keys and certs over DNS using DNSSEC as an alternative to the CA model.
 https://www.rfc-editor.org/rfc/rfc6698.txt

 An initial patch was posted by Suresh Krishnaswamy on March 7th 2013
 (https://curl.haxx.se/mail/lib-2013-03/0075.html) but it was a too simple
 approach. See Daniel's comments:
 https://curl.haxx.se/mail/lib-2013-03/0103.html . libunbound may be the
 correct library to base this development on.

 Björn Stenberg wrote a separate initial take on DANE that was never
 completed.

13.9 Configurable loading of OpenSSL configuration file

 libcurl calls the OpenSSL function CONF_modules_load_file() in openssl.c,
 Curl_ossl_init(). "We regard any changes in the OpenSSL configuration as a
 security risk or at least as unnecessary."

 Please add a configuration switch or something similar to disable the
 CONF_modules_load_file() call.

 See https://github.com/curl/curl/issues/2724

13.10 Support Authority Information Access certificate extension (AIA)

 AIA can provide various things like CRLs but more importantly information
 about intermediate CA certificates that can allow validation path to be
 fullfilled when the HTTPS server doesn't itself provide them.

 Since AIA is about downloading certs on demand to complete a TLS handshake,
 it is probably a bit tricky to get done right.

 See https://github.com/curl/curl/issues/2793

13.11 Support intermediate & root pinning for PINNEDPUBLICKEY

 CURLOPT_PINNEDPUBLICKEY does not consider the hashes of intermediate & root
 certificates when comparing the pinned keys. Therefore it is not compatible
 with "HTTP Public Key Pinning" as there also intermediate and root certificates
 can be pinned. This is very useful as it prevents webadmins from "locking
 themself out of their servers".

 Adding this feature would make curls pinning 100% compatible to HPKP and allow
 more flexible pinning.

13.12 Support HSTS

 "HTTP Strict Transport Security" is TOFU (trust on first use), time-based
 features indicated by a HTTP header send by the webserver. It is widely used
 in browsers and it's purpose is to prevent insecure HTTP connections after
 a previous HTTPS connection. It protects against SSLStripping attacks.

 Doc: https://developer.mozilla.org/en-US/docs/Web/Security/HTTP_strict_transport_security
 RFC 6797: https://tools.ietf.org/html/rfc6797

13.13 Support HPKP

 "HTTP Public Key Pinning" is TOFU (trust on first use), time-based
 features indicated by a HTTP header send by the webserver. It's purpose is
 to prevent Man-in-the-middle attacks by trusted CAs by allowing webadmins
 to specify which CAs/certificates/public keys to trust when connection to
 their websites.

 It can be build based on PINNEDPUBLICKEY.

 Wikipedia: https://en.wikipedia.org/wiki/HTTP_Public_Key_Pinning
 OWASP: https://www.owasp.org/index.php/Certificate_and_Public_Key_Pinning
 Doc: https://developer.mozilla.org/de/docs/Web/Security/Public_Key_Pinning
 RFC: https://tools.ietf.org/html/draft-ietf-websec-key-pinning-21

13.14 Support the clienthello extension

 Certain stupid networks and middle boxes have a problem with SSL handshake
 pakets that are within a certain size range because how that sets some bits
 that previously (in older TLS version) were not set. The clienthello
 extension adds padding to avoid that size range.

 https://tools.ietf.org/html/rfc7685
 https://github.com/curl/curl/issues/2299

14. GnuTLS

14.1 SSL engine stuff

 Is this even possible?

14.2 check connection

 Add a way to check if the connection seems to be alive, to correspond to the
 SSL_peak() way we use with OpenSSL.

15. WinSSL/SChannel

15.1 Add support for client certificate authentication

 WinSSL/SChannel currently makes use of the OS-level system and user
 certificate and private key stores. This does not allow the application
 or the user to supply a custom client certificate using curl or libcurl.

 Therefore support for the existing -E/--cert and --key options should be
 implemented by supplying a custom certificate to the SChannel APIs, see:
 - Getting a Certificate for Schannel
   https://msdn.microsoft.com/en-us/library/windows/desktop/aa375447.aspx

15.3 Add support for the --ciphers option

 The cipher suites used by WinSSL/SChannel are configured on an OS-level
 instead of an application-level. This does not allow the application or
 the user to customize the configured cipher suites using curl or libcurl.

 Therefore support for the existing --ciphers option should be implemented
 by mapping the OpenSSL/GnuTLS cipher suites to the SChannel APIs, see
 - Specifying Schannel Ciphers and Cipher Strengths
   https://msdn.microsoft.com/en-us/library/windows/desktop/aa380161.aspx

16. SASL

16.1 Other authentication mechanisms

 Add support for other authentication mechanisms such as OLP,
 GSS-SPNEGO and others.

16.2 Add QOP support to GSSAPI authentication

 Currently the GSSAPI authentication only supports the default QOP of auth
 (Authentication), whilst Kerberos V5 supports both auth-int (Authentication
 with integrity protection) and auth-conf (Authentication with integrity and
 privacy protection).

16.3 Support binary messages (i.e.: non-base64)

  Mandatory to support LDAP SASL authentication.


17. SSH protocols

17.1 Multiplexing

 SSH is a perfectly fine multiplexed protocols which would allow libcurl to do
 multiple parallel transfers from the same host using the same connection,
 much in the same spirit as HTTP/2 does. libcurl however does not take
 advantage of that ability but will instead always create a new connection for
 new transfers even if an existing connection already exists to the host.

 To fix this, libcurl would have to detect an existing connection and "attach"
 the new transfer to the existing one.

17.2 SFTP performance

 libcurl's SFTP transfer performance is sub par and can be improved, mostly by
 the approach mentioned in "1.6 Modified buffer size approach".

17.3 Support better than MD5 hostkey hash

 libcurl offers the CURLOPT_SSH_HOST_PUBLIC_KEY_MD5 option for verifying the
 server's key. MD5 is generally being deprecated so we should implement
 support for stronger hashing algorithms. libssh2 itself is what provides this
 underlying functionality and it supports at least SHA-1 as an alternative.
 SHA-1 is also being deprecated these days so we should consider working with
 libssh2 to instead offer support for SHA-256 or similar.

17.4 Support CURLOPT_PREQUOTE

 The two other QUOTE options are supported for SFTP, but this was left out for
 unknown reasons!

18. Command line tool

18.1 sync

 "curl --sync http://example.com/feed[1-100].rss" or
 "curl --sync http://example.net/{index,calendar,history}.html"

 Downloads a range or set of URLs using the remote name, but only if the
 remote file is newer than the local file. A Last-Modified HTTP date header
 should also be used to set the mod date on the downloaded file.

18.2 glob posts

 Globbing support for -d and -F, as in 'curl -d "name=foo[0-9]" URL'.
 This is easily scripted though.

18.3 prevent file overwriting

 Add an option that prevents curl from overwriting existing local files. When
 used, and there already is an existing file with the target file name
 (either -O or -o), a number should be appended (and increased if already
 existing). So that index.html becomes first index.html.1 and then
 index.html.2 etc.

18.4 simultaneous parallel transfers

 The client could be told to use maximum N simultaneous parallel transfers and
 then just make sure that happens. It should of course not make more than one
 connection to the same remote host. This would require the client to use the
 multi interface. https://curl.haxx.se/bug/feature.cgi?id=1558595

 Using the multi interface would also allow properly using parallel transfers
 with HTTP/2 and supporting HTTP/2 server push from the command line.

18.5 UTF-8 filenames in Content-Disposition

 RFC 6266 documents how UTF-8 names can be passed to a client in the
 Content-Disposition header, and curl does not support this.

 https://github.com/curl/curl/issues/1888

18.6 warning when setting an option

 Display a warning when libcurl returns an error when setting an option.
 This can be useful to tell when support for a particular feature hasn't been
 compiled into the library.

18.7 warning if curl version is not in sync with libcurl version

 This is usually a sign of a funny, weird or unexpected install situations
 that aren't always quickly nor easily detected by users. curl and libcurl are
 always released in sync and should use the same version numbers unless very
 special situations.

18.8 offer color-coded HTTP header output

 By offering different color output on the header name and the header
 contents, they could be made more readable and thus help users working on
 HTTP services.

18.9 Choose the name of file in braces for complex URLs

 When using braces to download a list of URLs and you use complicated names
 in the list of alternatives, it could be handy to allow curl to use other
 names when saving.

 Consider a way to offer that. Possibly like
 {partURL1:name1,partURL2:name2,partURL3:name3} where the name following the
 colon is the output name.

 See https://github.com/curl/curl/issues/221

18.10 improve how curl works in a windows console window

 If you pull the scrollbar when transferring with curl in a Windows console
 window, the transfer is interrupted and can get disconnected. This can
 probably be improved. See https://github.com/curl/curl/issues/322

18.11 -w output to stderr

 -w is quite useful, but not to those of us who use curl without -o or -O
 (such as for scripting through a higher level language). It would be nice to
 have an option that is exactly like -w but sends it to stderr
 instead. Proposed name: --write-stderr. See
 https://github.com/curl/curl/issues/613

18.12 keep running, read instructions from pipe/socket

 Provide an option that makes curl not exit after the last URL (or even work
 without a given URL), and then make it read instructions passed on a pipe or
 over a socket to make further instructions so that a second subsequent curl
 invoke can talk to the still running instance and ask for transfers to get
 done, and thus maintain its connection pool, DNS cache and more.

18.13 support metalink in http headers

 Curl has support for downloading a metalink xml file, processing it, and then
 downloading the target of the metalink. This is done via the --metalink option.
 It would be nice if metalink also supported downloading via metalink
 information that is stored in HTTP headers (RFC 6249). Theoretically this could
 also be supported with the --metalink option.

 See https://tools.ietf.org/html/rfc6249

 See also https://lists.gnu.org/archive/html/bug-wget/2015-06/msg00034.html for
 an implematation of this in wget.

18.14 --fail without --location should treat 3xx as a failure

 To allow a command line like this to detect a redirect and consider it a
 failure:

    curl -v --fail -O https://example.com/curl-7.48.0.tar.gz

 ... --fail must treat 3xx responses as failures too. The least problematic
 way to implement this is probably to add that new logic in the command line
 tool only and not in the underlying CURLOPT_FAILONERROR logic.

18.15 --retry should resume

 When --retry is used and curl actually retries transfer, it should use the
 already transferred data and do a resumed transfer for the rest (when
 possible) so that it doesn't have to transfer the same data again that was
 already transferred before the retry.

 See https://github.com/curl/curl/issues/1084

18.16 send only part of --data

 When the user only wants to send a small piece of the data provided with
 --data or --data-binary, like when that data is a huge file, consider a way
 to specify that curl should only send a piece of that. One suggested syntax
 would be: "--data-binary @largefile.zip!1073741823-2147483647".

 See https://github.com/curl/curl/issues/1200

18.17 consider file name from the redirected URL with -O ?

 When a user gives a URL and uses -O, and curl follows a redirect to a new
 URL, the file name is not extracted and used from the newly redirected-to URL
 even if the new URL may have a much more sensible file name.

 This is clearly documented and helps for security since there's no surprise
 to users which file name that might get overwritten. But maybe a new option
 could allow for this or maybe -J should imply such a treatment as well as -J
 already allows for the server to decide what file name to use so it already
 provides the "may overwrite any file" risk.

 This is extra tricky if the original URL has no file name part at all since
 then the current code path will error out with an error message, and we can't
 *know* already at that point if curl will be redirected to a URL that has a
 file name...

 See https://github.com/curl/curl/issues/1241

18.18 retry on network is unreachable

 The --retry option retries transfers on "transient failures". We later added
 --retry-connrefused to also retry for "connection refused" errors.

 Suggestions have been brought to also allow retry on "network is unreachable"
 errors and while totally reasonable, maybe we should consider a way to make
 this more configurable than to add a new option for every new error people
 want to retry for?

 https://github.com/curl/curl/issues/1603

18.19 expand ~/ in config files

 For example .curlrc could benefit from being able to do this.

 See https://github.com/curl/curl/issues/2317

18.20 host name sections in config files

 config files would be more powerful if they could set different
 configurations depending on used URLs, host name or possibly origin. Then a
 default .curlrc could a specific user-agent only when doing requests against
 a certain site.


19. Build

19.1 roffit

 Consider extending 'roffit' to produce decent ASCII output, and use that
 instead of (g)nroff when building src/tool_hugehelp.c

19.2 Enable PIE and RELRO by default

 Especially when having programs that execute curl via the command line, PIE
 renders the exploitation of memory corruption vulnerabilities a lot more
 difficult. This can be attributed to the additional information leaks being
 required to conduct a successful attack. RELRO, on the other hand, masks
 different binary sections like the GOT as read-only and thus kills a handful
 of techniques that come in handy when attackers are able to arbitrarily
 overwrite memory. A few tests showed that enabling these features had close
 to no impact, neither on the performance nor on the general functionality of
 curl.


20. Test suite

20.1 SSL tunnel

 Make our own version of stunnel for simple port forwarding to enable HTTPS
 and FTP-SSL tests without the stunnel dependency, and it could allow us to
 provide test tools built with either OpenSSL or GnuTLS

20.2 nicer lacking perl message

 If perl wasn't found by the configure script, don't attempt to run the tests
 but explain something nice why it doesn't.

20.3 more protocols supported

 Extend the test suite to include more protocols. The telnet could just do FTP
 or http operations (for which we have test servers).

20.4 more platforms supported

 Make the test suite work on more platforms. OpenBSD and Mac OS. Remove
 fork()s and it should become even more portable.

20.5 Add support for concurrent connections

 Tests 836, 882 and 938 were designed to verify that separate connections aren't
 used when using different login credentials in protocols that shouldn't re-use
 a connection under such circumstances.

 Unfortunately, ftpserver.pl doesn't appear to support multiple concurrent
 connections. The read while() loop seems to loop until it receives a disconnect
 from the client, where it then enters the waiting for connections loop. When
 the client opens a second connection to the server, the first connection hasn't
 been dropped (unless it has been forced - which we shouldn't do in these tests)
 and thus the wait for connections loop is never entered to receive the second
 connection.

20.6 Use the RFC6265 test suite

 A test suite made for HTTP cookies (RFC 6265) by Adam Barth is available at
 https://github.com/abarth/http-state/tree/master/tests

 It'd be really awesome if someone would write a script/setup that would run
 curl with that test suite and detect deviances. Ideally, that would even be
 incorporated into our regular test suite.


21. Next SONAME bump

21.1 http-style HEAD output for FTP

 #undef CURL_FTP_HTTPSTYLE_HEAD in lib/ftp.c to remove the HTTP-style headers
 from being output in NOBODY requests over FTP

21.2 combine error codes

 Combine some of the error codes to remove duplicates.  The original
 numbering should not be changed, and the old identifiers would be
 macroed to the new ones in an CURL_NO_OLDIES section to help with
 backward compatibility.

 Candidates for removal and their replacements:

    CURLE_FILE_COULDNT_READ_FILE => CURLE_REMOTE_FILE_NOT_FOUND

    CURLE_FTP_COULDNT_RETR_FILE => CURLE_REMOTE_FILE_NOT_FOUND

    CURLE_FTP_COULDNT_USE_REST => CURLE_RANGE_ERROR

    CURLE_FUNCTION_NOT_FOUND => CURLE_FAILED_INIT

    CURLE_LDAP_INVALID_URL => CURLE_URL_MALFORMAT

    CURLE_TFTP_NOSUCHUSER => CURLE_TFTP_ILLEGAL

    CURLE_TFTP_NOTFOUND => CURLE_REMOTE_FILE_NOT_FOUND

    CURLE_TFTP_PERM => CURLE_REMOTE_ACCESS_DENIED

21.3 extend CURLOPT_SOCKOPTFUNCTION prototype

 The current prototype only provides 'purpose' that tells what the
 connection/socket is for, but not any protocol or similar. It makes it hard
 for applications to differentiate on TCP vs UDP and even HTTP vs FTP and
 similar.

22. Next major release

22.1 cleanup return codes

 curl_easy_cleanup() returns void, but curl_multi_cleanup() returns a
 CURLMcode. These should be changed to be the same.

22.2 remove obsolete defines

 remove obsolete defines from curl/curl.h

22.3 size_t

 make several functions use size_t instead of int in their APIs

22.4 remove several functions

 remove the following functions from the public API:

 curl_getenv

 curl_mprintf (and variations)

 curl_strequal

 curl_strnequal

 They will instead become curlx_ - alternatives. That makes the curl app
 still capable of using them, by building with them from source.

 These functions have no purpose anymore:

 curl_multi_socket

 curl_multi_socket_all

22.5 remove CURLOPT_FAILONERROR

 Remove support for CURLOPT_FAILONERROR, it has gotten too kludgy and weird
 internally. Let the app judge success or not for itself.

22.6 remove CURLOPT_DNS_USE_GLOBAL_CACHE

 Remove support for a global DNS cache. Anything global is silly, and we
 already offer the share interface for the same functionality but done
 "right".

22.7 remove progress meter from libcurl

 The internally provided progress meter output doesn't belong in the library.
 Basically no application wants it (apart from curl) but instead applications
 can and should do their own progress meters using the progress callback.

 The progress callback should then be bumped as well to get proper 64bit
 variable types passed to it instead of doubles so that big files work
 correctly.

22.8 remove 'curl_httppost' from public

 curl_formadd() was made to fill in a public struct, but the fact that the
 struct is public is never really used by application for their own advantage
 but instead often restricts how the form functions can or can't be modified.

 Changing them to return a private handle will benefit the implementation and
 allow us much greater freedoms while still maintaining a solid API and ABI.