blob: c135caba5dfb64142beec1cd6a734d65337b688d [file] [log] [blame]
Andy Greena35c86f2013-01-31 10:16:44 +08001Changelog
2---------
3
Enno Boland7731a3e2016-05-05 22:08:41 +02004Fixes
5-----
6
71) OpenSSL version tests not needed on LibreSSL and BoringSSL
8
Enno Bolandb6e2ad62016-05-05 22:09:54 +020092) Fix IPV6 build breakage
10
OndraCo624b23d2016-05-05 12:57:11 +0200113) Some fixes for WinCE build
Enno Bolandb6e2ad62016-05-05 22:09:54 +020012
Andy Greenf4767492016-05-06 08:02:57 +0800134) Additional canned mimetypes for mounts, the full list is
14
15 .ico image/x-icon
16 .gif image/gif
17 .js text/javascript
18 .png image/png
19 .jpg image/jpeg
20 .gz application/gzip
21 .JPG image/jpeg
22 .html text/html
23 .css text/css
24 .txt text/plain
25 .ttf application/x-font-ttf
26 .woff application/font-woff
27 .xml application/xml
28
Andy Greenf6585282016-05-06 14:24:59 +0800295) Allow per-vhost setting of which protocol should get used
30when the protocol: header is not sent by the client
Andy Greenf4767492016-05-06 08:02:57 +080031
Enno Boland7731a3e2016-05-05 22:08:41 +020032
Andy Green55006432016-05-05 09:40:18 +080033v2.0.0
34======
35
36Summary
37-------
38
39 - There are only api additions, the api is compatible with v1.7.x. But
40 there is necessarily an soname bump to 8.
41
42 - If you are using lws client, you mainly need to be aware the option
43 LWS_SERVER_OPTION_DO_SSL_GLOBAL_INIT is needed at context-creation time
44 if you will use SSL.
45
46 - If you are using lws for serving, the above is also true but there are
47 many new features to simplify your code (and life). There is a
48 summany online here
49
50 https://libwebsockets.org/lws-2.0-new-features.html
51
52 but basically the keywords are vhosts, mounts and plugins. You can now
53 do the web serving part from lws without any user callback code at all.
54 See ./test-server/test-server-v2.0.c for an example, it has no user
55 code for ws either since it uses the protocol plugins... that one C file
56 is all that is needed to do the whole test server function.
57
58 You now have the option to use a small generic ws-capable webserver
59 "lwsws" and write your ws part as a plugin. That eliminates even
60 cut-and-pasting the test server code and offers more configurable
61 features like control over http cacheability in JSON.
62
63
Andy Green0c3cc2e2016-02-20 09:12:52 +080064Fixes
65-----
66
Andy Green55006432016-05-05 09:40:18 +080067These are already in 1.7.x series
68
Andy Green0c3cc2e2016-02-20 09:12:52 +0800691) MAJOR (Windows-only) fix assert firing
70
712) MAJOR http:/1.1 connections handled by lws_return_http_status() did not
72get sent a content-length resulting in the link hanging until the peer closed
73it. attack.sh updated to add a test for this.
74
Andy Green26d42492016-02-24 12:40:21 +0800753) MINOR An error about hdr struct in _lws_ws_related is corrected, it's not
76known to affect anything until after it was fixed
77
Andy Green5c0bcf42016-02-24 21:27:46 +0800784) MINOR During the close shutdown wait state introduced at v1.7, if something
79requests callback on writeable for the socket it will busywait until the
80socket closes
81
Andy Greendbfbbb42016-02-24 20:58:19 +0800825) MAJOR Although the test server has done it for a few versions already, it
83is now required for the user code to explicitly call
84
85 if (lws_http_transaction_completed(wsi))
86 return -1;
87
88when it finishes replying to a transaction in http. Previously the library
89did it for you, but that disallowed large, long transfers with multiple
90trips around the event loop (and cgi...).
91
Andy Green4ba798d2016-02-25 21:50:49 +0800926) MAJOR connections on ah waiting list that closed did not get removed from
93the waiting list...
94
Andy Green83af28a2016-02-28 10:55:31 +0800957) MAJOR since we added the ability to hold an ah across http keepalive
96transactions where more headers had already arrived, we broke the ability
97to tell if more headers had arrived. Result was if the browser didn't
98close the keepalive, we retained ah for the lifetime of the keepalive,
99using up the pool.
100
Andy Green442e1c82016-02-29 10:10:42 +08001018) MAJOR windows-only-POLLHUP was not coming
102
Andy Green4f5ebec2016-03-09 23:13:31 +08001039) Client should not send ext hdr if no exts
Andy Greendbfbbb42016-02-24 20:58:19 +0800104
Andy Green0c3cc2e2016-02-20 09:12:52 +0800105Changes
106-------
107
1081) MINOR test-server gained some new switches
109
110 -C <file> use external SSL cert file
111 -K <file> use external SSL key file
112 -A <file> use external SSL CA cert file
113
114 -u <uid> set effective uid
115 -g <gid> set effective gid
116
117together you can use them like this to have the test-server work with the
118usual purchased SSL certs from an official CA.
119
120 --ssl -C your.crt -K your.key -A your.cer -u 99 -g 99
121
1222) MINOR the OpenSSL magic to setup ECDH cipher usage is implemented in the
123library, and the ciphers restricted to use ECDH only.
124Using this, the lws test server can score an A at SSLLABS test
125
1263) MINOR STS (SSL always) header is added to the test server if you use --ssl. With
127that, we score A+ at SSLLABS test
128
1294) MINOR daemonize function (disabled at cmake by default) is updated to work
130with systemd
131
1325) MINOR example systemd .service file now provided for test server
133(not installed by default)
134
Andy Green0ad1a6e2016-02-20 14:05:55 +08001356) test server html is updated with tabs and a new live server monitoring
136feature. Input sanitization added to the js.
137
Andy Green2d8d35a2016-02-29 14:19:16 +08001387) client connections attempted when no ah is free no longer fail, they are
139just deferred until an ah becomes available.
140
Andy Greena661ee52016-02-29 13:18:30 +08001418) The test client pays attention to if you give it an http:/ or https://
142protocol string to its argument in URL format. If so, it stays in http[s]
143client mode and doesn't upgrade to ws[s], allowing you to do generic http client
Andy Green5c8906e2016-03-13 16:44:19 +0800144operations. Receiving transfer-encoding: chunked is supported.
Andy Greena661ee52016-02-29 13:18:30 +0800145
Andy Green1e5a9ad2016-03-20 11:59:53 +08001469) If you enable -DLWS_WITH_HTTP_PROXY=1 at cmake, the test server has a
147new URI path http://localhost:7681/proxytest If you visit here, a client
148connection to http://example.com:80 is spawned, and the results piped on
149to your original connection.
150
15110) Also with LWS_WITH_HTTP_PROXY enabled at cmake, lws wants to link to an
152additional library, "libhubbub". This allows lws to do html rewriting on the
153fly, adjusting proxied urls in a lightweight and fast way.
Andy Greenfb5f33b2016-03-01 07:19:01 +0800154
Andy Greenc6fd3602016-03-23 09:22:11 +080015511) There's a new context creation flag LWS_SERVER_OPTION_DO_SSL_GLOBAL_INIT,
156this is included automatically if you give any other SSL-related option flag.
157If you give no SSL-related option flag, nor this one directly, then even
158though SSL support may be compiled in, it is never initialized nor used for the
159whole lifetime of the lws context.
160
161Conversely in order to prepare the context to use SSL, even though, eg, you
162are not listening on SSL but will use SSL client connections later, you must
163give this flag explicitly to make sure SSL is initialized.
164
Andy Green2d8d35a2016-02-29 14:19:16 +0800165
Andy Greend7fddad2016-02-18 20:36:55 +0800166User API additions
167------------------
168
Andy Green0c3cc2e2016-02-20 09:12:52 +08001691) MINOR APIBREAK There's a new member in struct lws_context_creation_info, ecdh_curve,
Andy Greend7fddad2016-02-18 20:36:55 +0800170which lets you set the name of the ECDH curve OpenSSL should use. By
171default (if you leave ecdh_curve NULL) it will use "prime256v1"
172
Andy Green51d9afa2016-02-24 11:05:56 +08001732) MINOR NEWAPI It was already possible to adopt a foreign socket that had not
174been read from using lws_adopt_socket() since v1.7. Now you can adopt a
175partially-used socket if you don't need SSL, by passing it what you read
176so it can drain that before reading from the socket.
177
178LWS_VISIBLE LWS_EXTERN struct lws *
179lws_adopt_socket_readbuf(struct lws_context *context, lws_sockfd_type accept_fd,
180 const char *readbuf, size_t len);
Andy Greend7fddad2016-02-18 20:36:55 +0800181
Andy Green6a8099b2016-02-21 21:25:48 +08001823) MINOR NEWAPI CGI type "network io" subprocess execution is now possible from
183a simple api.
184
185LWS_VISIBLE LWS_EXTERN int
Andy Green1a138852016-03-20 11:55:25 +0800186lws_cgi(struct lws *wsi, char * const *exec_array, int script_uri_path_len,
187 int timeout_secs);
Andy Green6a8099b2016-02-21 21:25:48 +0800188
189LWS_VISIBLE LWS_EXTERN int
190lws_cgi_kill(struct lws *wsi);
191
192To use it, you must first set the cmake option
193
194$ cmake .. -DLWS_WITH_CGI=1
195
196See test-server-http.c and test server path
197
198http://localhost:7681/cgitest
199
200stdin gets http body, you can test it with wget
201
202$ echo hello > hello.txt
203$ wget http://localhost:7681/cgitest --post-file=hello.txt -O- --quiet
204lwstest script
205read="hello"
206
Andy Green1a138852016-03-20 11:55:25 +0800207The test script returns text/html table showing /proc/meminfo. But the cgi
208support is complete enough to run cgit cgi.
209
Andy Greend61bed32016-02-25 15:01:55 +08002104) There is a helper api for forming logging timestamps
211
212LWS_VISIBLE int
213lwsl_timestamp(int level, char *p, int len)
214
215this generates this kind of timestamp for use as logging preamble
216
217lwsts[13116]: [2016/01/25 14:52:52:8386] NOTICE: Initial logging level 7
218
Andy Greena661ee52016-02-29 13:18:30 +08002195) struct lws_client_connect_info has a new member
220
221 const char *method
222
223If it's NULL, then everything happens as before, lws_client_connect_via_info()
224makes a ws or wss connection to the address given.
225
226If you set method to a valid http method like "GET", though, then this method
227is used and the connection remains in http[s], it's not upgraded to ws[s].
228
229So with this, you can perform http[s] client operations as well as ws[s] ones.
230
231There are 4 new related callbacks
232
233 LWS_CALLBACK_ESTABLISHED_CLIENT_HTTP = 44,
234 LWS_CALLBACK_CLOSED_CLIENT_HTTP = 45,
235 LWS_CALLBACK_RECEIVE_CLIENT_HTTP = 46,
236 LWS_CALLBACK_COMPLETED_CLIENT_HTTP = 47,
Andy Green6a8099b2016-02-21 21:25:48 +0800237
Andy Green494418a2016-03-02 09:17:22 +08002386) struct lws_client_connect_info has a new member
239
240 const char *parent_wsi
241
242if non-NULL, the client wsi is set to be a child of parent_wsi. This ensures
243if parent_wsi closes, then the client child is closed just before.
244
Andy Green0f9904f2016-03-17 15:26:49 +08002457) If you're using SSL, there's a new context creation-time option flag
246LWS_SERVER_OPTION_REDIRECT_HTTP_TO_HTTPS. If you give this, non-ssl
247connections to the server listen port are accepted and receive a 301
248redirect to / on the same host and port using https://
249
Andy Greenc5376b12016-04-08 09:45:49 +08002508) User code may set per-connection extension options now, using a new api
251"lws_set_extension_option()".
252
253This should be called from the ESTABLISHED callback like this
254
255 lws_set_extension_option(wsi, "permessage-deflate",
256 "rx_buf_size", "12"); /* 1 << 12 */
257
258If the extension is not active (missing or not negotiated for the
259connection, or extensions are disabled on the library) the call is
260just returns -1. Otherwise the connection's extension has its
261named option changed.
262
263The extension may decide to alter or disallow the change, in the
264example above permessage-deflate restricts the size of his rx
265output buffer also considering the protocol's rx_buf_size member.
266
267
Andy Greencd0c6962016-03-28 10:12:37 +0800268New application lwsws
269---------------------
270
271A libwebsockets-based general webserver is built by default now, lwsws.
272
273It's configured by JSON, by default in
274
275 /etc/lwsws/conf
276
277which contains global lws context settings like this
278
279{
280 "global": {
281 "uid": "99",
282 "gid": "99",
283 "interface": "eth0",
284 "count-threads": "1"
285 }
286}
287
288 /etc/lwsws/conf.d/*
289
290which contains zero or more files describing vhosts, like this
291
292{
293 "vhosts": [
294 { "name": "warmcat.com",
295 "port": "443",
296 "host-ssl-key": "/etc/pki/tls/private/warmcat.com.key",
297 "host-ssl-cert": "/etc/pki/tls/certs/warmcat.com.crt",
298 "host-ssl-ca": "/etc/pki/tls/certs/warmcat.com.cer",
299 "mounts": [
300 { "/": [
301 { "home": "file:///var/www/warmcat.com" },
302 { "default": "index.html" }
303 ]
304 }
305 ]
306 }
307 ]
308}
309
310
Andy Green0c3cc2e2016-02-20 09:12:52 +0800311
Andy Green9c60ed92016-02-16 12:32:18 +0800312v1.7.0
313======
314
Andy Green67112662016-01-11 11:34:01 +0800315Extension Changes
316-----------------
317
3181) There is now a "permessage-deflate" / RFC7692 implementation. It's very
319similar to "deflate-frame" we have offered for a long while; deflate-frame is
320now provided as an alias of permessage-deflate.
321
322The main differences are that the new permessage-deflate implementation:
323
324 - properly performs streaming respecting input and output buffer limits. The
325 old deflate-frame implementation could only work on complete deflate input
326 and produce complete inflate output for each frame. The new implementation
327 only mallocs buffers at initialization.
328
329 - goes around the event loop after each input package is processed allowing
330 interleaved output processing. The RX flow control api can be used to
331 force compressed input processing to match the rate of compressed output
332 processing (test--echo shows an example of how to do this).
333
334 - when being "deflate-frame" for compatibility he uses the same default zlib
335 settings as the old "deflate-frame", but instead of exponentially increasing
336 malloc allocations until the whole output will fit, he observes the default
337 input and output chunking buffer sizes of "permessage-deflate", that's
338 1024 in and 1024 out at a time.
339
3402) deflate-stream has been disabled for many versions (for over a year) and is
341now removed. Browsers are now standardizing on "permessage-deflate" / RFC7692
342
3433) struct lws_extension is simplified, and lws extensions now have a public
344api (their callback) for use in user code to compose extensions and options
345the user code wants. lws_get_internal_exts() is deprecated but kept around
346as a NOP. The changes allow one extension implementation to go by different
347names and allows the user client code to control option offers per-ext.
348
349The test client and server are updated to use the new way. If you use
350the old way it should still work, but extensions will be disabled until you
351update your code.
352
353Extensions are now responsible for allocating and per-instance private struct
354at instance construction time and freeing it when the instance is destroyed.
355Not needing to know the size means the extension's struct can be opaque
356to user code.
357
358
Andy Green3df58002015-12-25 12:44:12 +0800359User api additions
360------------------
361
Andy Green200a6a22016-02-15 20:36:02 +08003621) The info struct gained three new members
Andy Green3df58002015-12-25 12:44:12 +0800363
364 - max_http_header_data: 0 for default (1024) or set the maximum amount of known
365 http header payload that lws can deal with. Payload in unknown http
366 headers is dropped silently. If for some reason you need to send huge
367 cookies or other HTTP-level headers, you can now increase this at context-
368 creation time.
369
370 - max_http_header_pool: 0 for default (16) or set the maximum amount of http
371 headers that can be tracked by lws in this context. For the server, if
372 the header pool is completely in use then accepts on the listen socket
373 are disabled until one becomes free. For the client, if you simultaneously
374 have pending connects for more than this number of client connections,
375 additional connects will fail until some of the pending connections timeout
376 or complete.
377
Andy Green200a6a22016-02-15 20:36:02 +0800378 - timeout_secs: 0 for default (currently 20s), or set the library's
379 network activity timeout to the given number of seconds
380
Andy Green3df58002015-12-25 12:44:12 +0800381HTTP header processing in lws only exists until just after the first main
382callback after the HTTP handshake... for ws connections that is ESTABLISHED and
383for HTTP connections the HTTP callback.
384
385So these settings are not related to the maximum number of simultaneous
Andy Green3246ebb2015-12-26 12:03:06 +0800386connections, but the number of HTTP handshakes that may be expected or ongoing,
Andy Green3df58002015-12-25 12:44:12 +0800387or have just completed, at one time. The reason it's useful is it changes the
388memory allocation for header processing to be one-time at context creation
389instead of every time there is a new connection, and gives you control over
390the peak allocation.
391
392Setting max_http_header_pool to 1 is fine it will just queue incoming
393connections before the accept as necessary, you can still have as many
Andy Green5b85e392015-12-26 13:23:11 +0800394simultaneous post-header connections as you like. Since the http header
395processing is completed and the allocation released after ESTABLISHED or the
396HTTP callback, even with a pool of 1 many connections can be handled rapidly.
397
Andy Green066a7a12015-12-26 15:47:06 +08003982) There is a new callback that allows the user code to get acccess to the
399optional close code + aux data that may have been sent by the peer.
400
401LWS_CALLBACK_WS_PEER_INITIATED_CLOSE:
402 The peer has sent an unsolicited Close WS packet. @in and
403 @len are the optional close code (first 2 bytes, network
404 order) and the optional additional information which is not
405 defined in the standard, and may be a string or non-human-
406 readble data.
407 If you return 0 lws will echo the close and then close the
Andy Green67112662016-01-11 11:34:01 +0800408 connection. If you return nonzero lws will just close the
409 connection.
Andy Green066a7a12015-12-26 15:47:06 +0800410
411As usual not handling it does the right thing, if you're not interested in it
412just ignore it.
413
414The test server has "open and close" testing buttons at the bottom, if you
415open and close that connection, on close it will send a close code 3000 decimal
416and the string "Bye!" as the aux data.
417
418The test server dumb-increment callback handles this callback reason and prints
419
420lwsts[15714]: LWS_CALLBACK_WS_PEER_INITIATED_CLOSE: len 6
421lwsts[15714]: 0: 0x0B
422lwsts[15714]: 1: 0xB8
423lwsts[15714]: 2: 0x42
424lwsts[15714]: 3: 0x79
425lwsts[15714]: 4: 0x65
426lwsts[15714]: 5: 0x21
427
Andy Green1fb95e82015-12-26 17:20:34 +08004283) There is a new API to allow the user code to control the content of the
429close frame sent when about to return nonzero from the user callback to
430indicate the connection should close.
431
432/**
433 * lws_close_reason - Set reason and aux data to send with Close packet
434 * If you are going to return nonzero from the callback
435 * requesting the connection to close, you can optionally
436 * call this to set the reason the peer will be told if
437 * possible.
438 *
439 * @wsi: The websocket connection to set the close reason on
440 * @status: A valid close status from websocket standard
441 * @buf: NULL or buffer containing up to 124 bytes of auxiliary data
442 * @len: Length of data in @buf to send
443 */
444LWS_VISIBLE LWS_EXTERN void
445lws_close_reason(struct lws *wsi, enum lws_close_status status,
446 unsigned char *buf, size_t len);
447
448An extra button is added to the "open and close" test server page that requests
449that the test server close the connection from his end.
450
451The test server code will do so by
452
453 lws_close_reason(wsi, LWS_CLOSE_STATUS_GOINGAWAY,
454 (unsigned char *)"seeya", 5);
455 return -1;
456
457The browser shows the close code and reason he received
458
459websocket connection CLOSED, code: 1001, reason: seeya
460
Vijay Khurdiyae73d4462016-01-15 16:21:51 +08004614) There's a new context creation time option flag
Andy Green44a7f652015-12-29 11:20:09 +0800462
463LWS_SERVER_OPTION_VALIDATE_UTF8
464
465if you set it in info->options, then TEXT and CLOSE frames will get checked to
466confirm that they contain valid UTF-8. If they don't, the connection will get
467closed by lws.
468
Vijay Khurdiyae73d4462016-01-15 16:21:51 +08004695) ECDH Certs are now supported. Enable the CMake option
470
471cmake .. -DLWS_SSL_SERVER_WITH_ECDH_CERT=1
472
473**and** the info->options flag
474
Andy Greenff696482016-02-17 07:46:27 +0800475LWS_SERVER_OPTION_SSL_ECDH
Vijay Khurdiyae73d4462016-01-15 16:21:51 +0800476
477to build in support and select it at runtime.
478
Andy Greenff696482016-02-17 07:46:27 +08004796) There's a new api lws_parse_uri() that simplifies chopping up
Andy Green809d69a2016-01-14 11:37:56 +0800480https://xxx:yyy/zzz uris into parts nicely. The test client now uses this
Andy Greend3a55052016-01-19 03:34:24 +0800481to allow proper uris as well as the old address style.
482
4837) SMP support is integrated into LWS without any internal threading. It's
484very simple to use, libwebsockets-test-server-pthread shows how to do it,
485use -j <n> argument there to control the number of service threads up to 32.
486
487Two new members are added to the info struct
488
489 unsigned int count_threads;
490 unsigned int fd_limit_per_thread;
491
492leave them at the default 0 to get the normal singlethreaded service loop.
493
494Set count_threads to n to tell lws you will have n simultaneous service threads
495operating on the context.
496
497There is still a single listen socket on one port, no matter how many
498service threads.
499
500When a connection is made, it is accepted by the service thread with the least
501connections active to perform load balancing.
502
503The user code is responsible for spawning n threads running the service loop
504associated to a specific tsi (Thread Service Index, 0 .. n - 1). See
505the libwebsockets-test-server-pthread for how to do.
506
507If you leave fd_limit_per_thread at 0, then the process limit of fds is shared
508between the service threads; if you process was allowed 1024 fds overall then
509each thread is limited to 1024 / n.
510
511You can set fd_limit_per_thread to a nonzero number to control this manually, eg
512the overall supported fd limit is less than the process allowance.
513
514You can control the context basic data allocation for multithreading from Cmake
515using -DLWS_MAX_SMP=, if not given it's set to 32. The serv_buf allocation
516for the threads (currently 4096) is made at runtime only for active threads.
517
518Because lws will limit the requested number of actual threads supported
519according to LWS_MAX_SMP, there is an api lws_get_count_threads(context) to
520discover how many threads were actually allowed when the context was created.
521
522It's required to implement locking in the user code in the same way that
523libwebsockets-test-server-pthread does it, for the FD locking callbacks.
524
Andy Greenba119e92016-01-26 21:40:32 +0800525If LWS_MAX_SMP=1, then there is no code related to pthreads compiled in the
526library. If more than 1, a small amount of pthread mutex code is built into
527the library.
Andy Green809d69a2016-01-14 11:37:56 +0800528
Andy Greenba119e92016-01-26 21:40:32 +08005298) New API
530
531LWS_VISIBLE struct lws *
532lws_adopt_socket(struct lws_context *context, lws_sockfd_type accept_fd)
533
534allows foreign sockets accepted by non-lws code to be adopted by lws as if they
535had just been accepted by lws' own listen socket.
Andy Green3df58002015-12-25 12:44:12 +0800536
Danomi Czaski4e9c7f32016-01-28 09:40:53 +08005379) X-Real-IP: header has been added as WSI_TOKEN_HTTP_X_REAL_IP
538
Andy Green86ed65f2016-02-14 09:27:41 +080053910) Libuv support is added, there are new related user apis
540
541typedef void (lws_uv_signal_cb_t)(uv_loop_t *l, uv_signal_t *w, int revents);
542
543LWS_VISIBLE LWS_EXTERN int
544lws_uv_sigint_cfg(struct lws_context *context, int use_uv_sigint,
545 lws_uv_signal_cb_t *cb);
546
547LWS_VISIBLE LWS_EXTERN int
548lws_uv_initloop(struct lws_context *context, uv_loop_t *loop, int tsi);
549
550LWS_VISIBLE void
551lws_uv_sigint_cb(uv_loop_t *loop, uv_signal_t *watcher, int revents);
552
553and CMAKE option
554
555LWS_WITH_LIBUV
556
Danomi Czaski4e9c7f32016-01-28 09:40:53 +0800557
Andy Green3246ebb2015-12-26 12:03:06 +0800558User api changes
559----------------
560
5611) LWS_SEND_BUFFER_POST_PADDING is now 0 and deprecated. You can remove it; if
562you still use it, obviously it does nothing. Old binary code with nonzero
563LWS_SEND_BUFFER_POST_PADDING is perfectly compatible, the old code just
Andy Green5b85e392015-12-26 13:23:11 +0800564allocated a buffer bigger than the library is going to use.
Andy Green3246ebb2015-12-26 12:03:06 +0800565
Andy Green5b85e392015-12-26 13:23:11 +0800566The example apps no longer use LWS_SEND_BUFFER_POST_PADDING.
Andy Green3246ebb2015-12-26 12:03:06 +0800567
Andy Green4e2ac762015-12-26 20:26:11 +0800568The only path who made use of it was sending with LWS_WRITE_CLOSE --->
Andy Green3246ebb2015-12-26 12:03:06 +0800569
Andy Green1fb95e82015-12-26 17:20:34 +08005702) Because of lws_close_reason() formalizing handling close frames,
571LWS_WRITE_CLOSE is removed from libwebsockets.h. It was only of use to send
572close frames...close frame content should be managed using lws_close_reason()
573now.
574
Andy Green44a7f652015-12-29 11:20:09 +08005753) We check for invalid CLOSE codes and complain about protocol violation in
576our close code. But it changes little since we were in the middle of closing
577anyway.
578
5794) zero-length RX frames and zero length TX frames are now allowed.
580
5815) Pings and close used to be limited to 124 bytes, the correct limit is 125
582so that is now also allowed.
Andy Green3246ebb2015-12-26 12:03:06 +0800583
Steffen Vogelf9267172016-02-09 07:19:15 +01005846) LWS_PRE is provided as a synonym for LWS_SEND_BUFFER_PRE_PADDING, either is
Andy Green67112662016-01-11 11:34:01 +0800585valid to use now.
586
5877) There's generic support for RFC7462 style extension options built into the
588library now. As a consequence, a field "options" is added to lws_extension.
589It can be NULL if there are no options on the extension. Extension internal
590info is part of the public abi because extensions may be implemented outside
591the library.
592
Danomi Czaski4e9c7f32016-01-28 09:40:53 +08005938) WSI_TOKEN_PROXY enum was accidentally defined to collide with another token
594of value 73. That's now corrected and WSI_TOKEN_PROXY moved to his own place at
59577.
596
Andy Green86ed65f2016-02-14 09:27:41 +08005979) With the addition of libuv support, libev is not the only event loop
598library in town and his api names must be elaborated with _ev_
599
600 Callback typedef: lws_signal_cb ---> lws_ev_signal_cb_t
601 lws_sigint_cfg --> lws_ev_sigint_cfg
602 lws_initloop --> lws_ev_initloop
603 lws_sigint_cb --> lws_ev_sigint_cb
604
60510) Libev support is made compatible with multithreaded service,
606lws_ev_initloop (was lws_initloop) gets an extra argument for the
607thread service index (use 0 if you will just have 1 service thread).
608
609LWS_VISIBLE LWS_EXTERN int
Andy Greenff696482016-02-17 07:46:27 +0800610lws_ev_initloop(struct lws_context *context, ev_loop_t *loop, int tsi);
Andy Green86ed65f2016-02-14 09:27:41 +0800611
Andy Green3df58002015-12-25 12:44:12 +0800612
Andy Greenaef3dc42016-05-06 07:45:19 +0800613(for earlier changelogs, see the tagged releases)