posix_mq 0.5.1 → 0.6.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
data/.document CHANGED
@@ -5,3 +5,4 @@ ChangeLog
5
5
  lib
6
6
  ext/posix_mq/posix_mq.c
7
7
  posix-mq-rb_1
8
+ LATEST
data/.gitignore CHANGED
@@ -12,3 +12,4 @@ Makefile
12
12
  /man
13
13
  /pkg
14
14
  /doc
15
+ /LATEST
data/.manifest CHANGED
@@ -1,6 +1,7 @@
1
1
  .document
2
2
  .gitignore
3
3
  .manifest
4
+ .wrongdoc.yml
4
5
  COPYING
5
6
  ChangeLog
6
7
  Documentation/.gitignore
@@ -9,6 +10,7 @@ Documentation/posix-mq-rb.1.txt
9
10
  GIT-VERSION-FILE
10
11
  GIT-VERSION-GEN
11
12
  GNUmakefile
13
+ LATEST
12
14
  LICENSE
13
15
  NEWS
14
16
  README
@@ -17,7 +19,6 @@ bin/posix-mq-rb
17
19
  ext/posix_mq/extconf.rb
18
20
  ext/posix_mq/posix_mq.c
19
21
  lib/posix_mq.rb
20
- local.mk.sample
21
22
  man/man1/posix-mq-rb.1
22
23
  posix_mq.gemspec
23
24
  setup.rb
data/.wrongdoc.yml ADDED
@@ -0,0 +1,7 @@
1
+ ---
2
+ rdoc_url: http://bogomips.org/ruby_posix_mq/
3
+ cgit_url: http://git.bogomips.org/cgit/ruby_posix_mq.git
4
+ git_url: git://git.bogomips.org/ruby_posix_mq.git
5
+ changelog_since: 0.4.0
6
+ merge_html:
7
+ posix-mq-rb_1: Documentation/posix-mq.rb.1.html
data/ChangeLog CHANGED
@@ -1,72 +1,502 @@
1
- ChangeLog from git://git.bogomips.org/ruby_posix_mq.git (v0.4.0..v0.5.1)
1
+ ChangeLog from http://git.bogomips.org/cgit/ruby_posix_mq.git
2
2
 
3
- commit 3ca83d322486bb3da20d974af7faa74f9df891d8
4
- Author: Eric Wong <normalperson@yhbt.net>
5
- Date: Sun May 9 01:05:56 2010 -0700
6
-
7
- posix_mq 0.5.1
8
-
9
- Fix POSIX_MQ#notify(&block) usage, this regression was
10
- introduced in 0.4.0 and our tests for it were broken, as well.
11
-
12
- commit a997f4822a99590c7a5175be4a694b4482a4b997
13
- Author: Eric Wong <normalperson@yhbt.net>
14
- Date: Sun May 9 00:05:03 2010 -0700
15
-
16
- fix POSIX_MQ#notify(&block) aka SIGEV_THREAD
17
-
18
- tests for them were stupidly broken and never executed :x
19
-
20
- commit f3605c820fd73713e34950170bf759e1af204038
21
- Author: Eric Wong <normalperson@yhbt.net>
22
- Date: Tue May 4 19:48:18 2010 -0700
23
-
24
- posix_mq 0.5.0
25
-
26
- Uncommonly raised exceptions due to programmer error are now
27
- raised more safely with respect to the MRI garbage collector.
28
-
29
- The "posix-mq.rb" executable has been renamed to "posix-mq-rb"
30
- to respect RPS and is also hopefully easier to type/read when
31
- used in scripts. I'm preserving the "rb" in the name since I
32
- hope to have a more portable, C-only version someday.
33
-
34
- Minor cleanups and documentation now uses RDoc 2.5.x
35
-
36
- commit f4d7d9d14cf65595aa3d98429a6732bd6b007b4a
37
- Author: Eric Wong <normalperson@yhbt.net>
38
- Date: Tue May 4 19:37:37 2010 -0700
39
-
40
- doc: updates for RDoc 2.5.x
41
-
42
- commit 7b1b8a1cbbedc2764d4a8ab0d992f9f2ef73d9b7
43
- Author: Eric Wong <normalperson@yhbt.net>
44
- Date: Tue May 4 19:31:47 2010 -0700
45
-
46
- rename `posix-mq.rb' executable to `posix-mq-rb'
47
-
48
- ".rb" in the executable name is discouraged by the
49
- {Ruby packaging standard}[http://github.com/chneukirchen/rps]
50
- and probably confusing.
51
-
52
- If I ever get around to it, a non-Ruby version of this
53
- executable is planned, and that will be called "posix-mq"
54
-
55
- commit 33368c839dc8c63489591bda6e60c78025be8852
56
- Author: Eric Wong <normalperson@yhbt.net>
57
- Date: Thu Apr 22 00:02:25 2010 -0700
58
-
59
- GC safety fixes
60
-
61
- Avoid the RSTRING_PTR(rb_inspect(val)) construct, as it may not
62
- preserve the value of rb_inspect() on the stack.
63
-
64
- commit a9a365326ea0f7a37852d25adfb03049c0b24c22
65
- Author: Eric Wong <normalperson@yhbt.net>
66
- Date: Wed Apr 21 23:58:25 2010 -0700
67
-
68
- cleaner lookup "Signal" of constant
69
-
70
- Instead of blindly defining a module, this is more flexible in
71
- case Signal in Ruby itself becomes something other than a
72
- module.
3
+ commit 88e85b58f3cb0796a680c86d83c16c305c33b098
4
+ Author: Eric Wong <normalperson@yhbt.net>
5
+ Date: Sat Dec 25 00:23:03 2010 -0800
6
+
7
+ posix_mq 0.6.0 - Rubinius support
8
+
9
+ Rubinius 1.2 as well as Ruby 1.9. Ruby 1.8 works except it will
10
+ block the entire interpreter for blocking operations. Spurious
11
+ errors during GC for POSIX_MQ#to_io users are now avoided.
12
+
13
+ commit 2384ecce25e91feb1ec0a9537c76beb18a7d4f86
14
+ Author: Eric Wong <normalperson@yhbt.net>
15
+ Date: Sat Dec 25 00:19:49 2010 -0800
16
+
17
+ switch to wrongdoc for documentation
18
+
19
+ No more JavaScript on our site!
20
+
21
+ commit 5339d9d693c794c0f20270d5726cb360e31dff30
22
+ Author: Eric Wong <normalperson@yhbt.net>
23
+ Date: Fri Dec 24 23:56:07 2010 -0800
24
+
25
+ add Rubinius support
26
+
27
+ Cleaner code, too, no more direct RSTRUCT usage.
28
+
29
+ commit 79583aaa426ac2f0967cb84524b6e17d72499cd5
30
+ Author: Eric Wong <normalperson@yhbt.net>
31
+ Date: Fri Dec 24 23:24:29 2010 -0800
32
+
33
+ refactor type checks and remove unnecessary guards
34
+
35
+ StringValuePtr should already be a sufficient guard.
36
+ There are more tests while we're at it, too, and we'll
37
+ now raise TypeError instead of ArgumentError when a
38
+ POSIX_MQ::Attr is not passed properly.
39
+
40
+ commit 31dee76d5ef41ec5524c3e4dfb802db4a7a468ef
41
+ Author: Eric Wong <normalperson@yhbt.net>
42
+ Date: Fri Dec 24 18:11:04 2010 -0800
43
+
44
+ remove rb_time_interval() and gettimeofday() dependency
45
+
46
+ This function isn't exported in the standard Ruby headers,
47
+ it returns an aggregate value and isn't available in Rubinius,
48
+ either, so nuke it.
49
+
50
+ While we're at it, use clock_gettime() instead of gettimeofday()
51
+ to avoid unnecessary timeval usage since mq_send/mq_receive
52
+ rely on higher-precision timespecs instead.
53
+
54
+ commit 748d4aa2d3ece18db7d48c355ce77c877e4d3bb9
55
+ Author: Eric Wong <normalperson@yhbt.net>
56
+ Date: Thu Aug 19 19:37:32 2010 -0700
57
+
58
+ tests: use Class#method_defined?
59
+
60
+ I did not know about this method back in the day...
61
+
62
+ commit 8f5be0c822362e113919145d52b9eeebba711236
63
+ Author: Eric Wong <normalperson@yhbt.net>
64
+ Date: Thu Aug 19 19:36:23 2010 -0700
65
+
66
+ test: avoid leaving queues behind
67
+
68
+ We shouldn't unnecessarily hog system resources.
69
+
70
+ commit cea38abf1f76b5b795b9341406ee814303f5752b
71
+ Author: Eric Wong <normalperson@yhbt.net>
72
+ Date: Thu Aug 19 19:25:58 2010 -0700
73
+
74
+ avoid double close() and EBADF
75
+
76
+ It can be dangerous to hit (and ignore) EBADF errors in
77
+ multi-threaded applications. Users of POSIX_MQ#to_io have two
78
+ Ruby objects pointing to the same file descriptor, making
79
+ things tricky when it comes time to reap resources.
80
+
81
+ We'll always prefer to close the Ruby IO object if it exists
82
+ (because we have less control over its GC behavior) and
83
+ ignore the raw descriptor.
84
+
85
+ commit 3cf7ec34b6319693af14215ebb0da51c3e6a4603
86
+ Author: Eric Wong <normalperson@yhbt.net>
87
+ Date: Mon Aug 2 08:28:15 2010 +0000
88
+
89
+ open/notify: invoke GC if needed
90
+
91
+ On ENOMEM, EMFILE, and ENFILE errors, it is customary to
92
+ invoke the Ruby GC and free up resources and retry the
93
+ system call.
94
+
95
+ commit 3ca83d322486bb3da20d974af7faa74f9df891d8
96
+ Author: Eric Wong <normalperson@yhbt.net>
97
+ Date: Sun May 9 01:05:56 2010 -0700
98
+
99
+ posix_mq 0.5.1
100
+
101
+ Fix POSIX_MQ#notify(&block) usage, this regression was
102
+ introduced in 0.4.0 and our tests for it were broken, as well.
103
+
104
+ commit a997f4822a99590c7a5175be4a694b4482a4b997
105
+ Author: Eric Wong <normalperson@yhbt.net>
106
+ Date: Sun May 9 00:05:03 2010 -0700
107
+
108
+ fix POSIX_MQ#notify(&block) aka SIGEV_THREAD
109
+
110
+ tests for them were stupidly broken and never executed :x
111
+
112
+ commit f3605c820fd73713e34950170bf759e1af204038
113
+ Author: Eric Wong <normalperson@yhbt.net>
114
+ Date: Tue May 4 19:48:18 2010 -0700
115
+
116
+ posix_mq 0.5.0
117
+
118
+ Uncommonly raised exceptions due to programmer error are now
119
+ raised more safely with respect to the MRI garbage collector.
120
+
121
+ The "posix-mq.rb" executable has been renamed to "posix-mq-rb"
122
+ to respect RPS and is also hopefully easier to type/read when
123
+ used in scripts. I'm preserving the "rb" in the name since I
124
+ hope to have a more portable, C-only version someday.
125
+
126
+ Minor cleanups and documentation now uses RDoc 2.5.x
127
+
128
+ commit f4d7d9d14cf65595aa3d98429a6732bd6b007b4a
129
+ Author: Eric Wong <normalperson@yhbt.net>
130
+ Date: Tue May 4 19:37:37 2010 -0700
131
+
132
+ doc: updates for RDoc 2.5.x
133
+
134
+ commit 7b1b8a1cbbedc2764d4a8ab0d992f9f2ef73d9b7
135
+ Author: Eric Wong <normalperson@yhbt.net>
136
+ Date: Tue May 4 19:31:47 2010 -0700
137
+
138
+ rename `posix-mq.rb' executable to `posix-mq-rb'
139
+
140
+ ".rb" in the executable name is discouraged by the
141
+ {Ruby packaging standard}[http://github.com/chneukirchen/rps]
142
+ and probably confusing.
143
+
144
+ If I ever get around to it, a non-Ruby version of this
145
+ executable is planned, and that will be called "posix-mq"
146
+
147
+ commit 33368c839dc8c63489591bda6e60c78025be8852
148
+ Author: Eric Wong <normalperson@yhbt.net>
149
+ Date: Thu Apr 22 00:02:25 2010 -0700
150
+
151
+ GC safety fixes
152
+
153
+ Avoid the RSTRING_PTR(rb_inspect(val)) construct, as it may not
154
+ preserve the value of rb_inspect() on the stack.
155
+
156
+ commit a9a365326ea0f7a37852d25adfb03049c0b24c22
157
+ Author: Eric Wong <normalperson@yhbt.net>
158
+ Date: Wed Apr 21 23:58:25 2010 -0700
159
+
160
+ cleaner lookup "Signal" of constant
161
+
162
+ Instead of blindly defining a module, this is more flexible in
163
+ case Signal in Ruby itself becomes something other than a
164
+ module.
165
+
166
+ commit aa9beecf64739ea084158ff63a92f626bfb5717d
167
+ Author: Eric Wong <normalperson@yhbt.net>
168
+ Date: Sat Mar 13 00:28:14 2010 -0800
169
+
170
+ posix_mq 0.4.0
171
+
172
+ Small bugfixes and small API changes to avoid potential
173
+ issues/misuse are the focus of this release.
174
+
175
+ For non-blocking operation, the GVL is no longer bounced. This
176
+ reduces synchronization/scheduling overhead when used in
177
+ non-blocking applications.
178
+
179
+ Small cleanups and documentation improvements, too.
180
+
181
+ * make POSIX_MQ#dup and POSIX_MQ#clone no-op
182
+ * do not release GVL for non-blocking operations
183
+ * do not release GVL when unlinking/opening
184
+ * POSIX_MQ#<< does not release GVL when non-blocking
185
+ * avoid shadow warnings
186
+ * README: add mailing list archives info
187
+ * POSIX_MQ#to_io works under FreeBSD, too
188
+ * fix potential race with notify(&block)
189
+ * add TODO item for using netlink under Linux
190
+ * remove non-portable #warning CPP directive
191
+ * ensure POSIX_MQ#name is clobber-proof
192
+
193
+ commit 437cbe78cb0f239e34f9dbf1341f5b85f96f2f25
194
+ Author: Eric Wong <normalperson@yhbt.net>
195
+ Date: Sat Mar 13 00:22:22 2010 -0800
196
+
197
+ ensure POSIX_MQ#name is clobber-proof
198
+
199
+ We don't want folks to accidentally clobber the value for
200
+ others, so allocate a new string object for it (don't worry,
201
+ rb_str_dup() is cheap in 1.9).
202
+
203
+ commit 74e1f75b238416d9ac402f291431a5e457ae313f
204
+ Author: Eric Wong <normalperson@yhbt.net>
205
+ Date: Sat Mar 13 00:08:37 2010 -0800
206
+
207
+ remove non-portable #warning CPP directive
208
+
209
+ There's a good chance they're not using GCC if they're on some
210
+ weird platform that we'd emit a warning for.
211
+
212
+ commit 9881f136858ac8f3a77a6cea8ce15b1cdb4ad55a
213
+ Author: Eric Wong <normalperson@yhbt.net>
214
+ Date: Tue Mar 2 22:47:29 2010 -0800
215
+
216
+ add TODO item for using netlink under Linux
217
+
218
+ commit 19668b0edc9b09cfd72247f7041390e0328032a0
219
+ Author: Eric Wong <normalperson@yhbt.net>
220
+ Date: Tue Mar 2 21:20:28 2010 -0800
221
+
222
+ fix potential race with notify(&block)
223
+
224
+ We need to assign the notify_thread before assigning the
225
+ notification. Otherwise, there's a chance the notification
226
+ could fire and the notify_thread is not properly assigned for
227
+ the POSIX_MQ object when the pipe becomes readable.
228
+
229
+ commit 0f4a8ef81479f9eff9037011d81b96cf402cbe7b
230
+ Author: Eric Wong <normalperson@yhbt.net>
231
+ Date: Sat Feb 20 17:36:16 2010 -0800
232
+
233
+ POSIX_MQ#to_io works under FreeBSD, too
234
+
235
+ commit 44cd880c96a76d43f6dcc7f7cdf7f687e32bda0c
236
+ Author: Eric Wong <normalperson@yhbt.net>
237
+ Date: Sat Feb 20 17:35:08 2010 -0800
238
+
239
+ README: add mailing list archives info
240
+
241
+ commit 4b56e475483454cd6616dd3c310c251940e9368b
242
+ Author: Eric Wong <normalperson@yhbt.net>
243
+ Date: Sat Feb 20 17:19:52 2010 -0800
244
+
245
+ avoid shadow warnings
246
+
247
+ commit 97ef1cfd699640e40d2d63746d005f4f4cf6a863
248
+ Author: Eric Wong <normalperson@yhbt.net>
249
+ Date: Sat Feb 20 14:52:16 2010 -0800
250
+
251
+ POSIX_MQ#<< does not release GVL when non-blocking
252
+
253
+ Missed this with the other change
254
+
255
+ commit 5849a3ca8e0691a7f39c5cdbfc09fc34ee4da308
256
+ Author: Eric Wong <normalperson@yhbt.net>
257
+ Date: Sat Feb 20 14:46:04 2010 -0800
258
+
259
+ do not release GVL when unlinking/opening
260
+
261
+ Since the message queue is not actually on a (slow) block
262
+ device, it's unlikely to block in a way where other tasks may be
263
+ scheduled by the kernel. So avoid complicating things and
264
+ unnecessary task switching and assume mq_open/mq_unlink can
265
+ be executed as fast as the CPU/memory subsystems allows.
266
+
267
+ commit a90d0f97c9eb5d9683af09b60ef7af9e94a255c3
268
+ Author: Eric Wong <normalperson@yhbt.net>
269
+ Date: Sat Feb 20 14:40:16 2010 -0800
270
+
271
+ do not release GVL for non-blocking operations
272
+
273
+ There's no point in wasting cycles releasing and reacquiring
274
+ a lock when we know we won't block. Since most non-blocking
275
+ users are expected to be single/few-threaded processes, this
276
+ will likely help them.
277
+
278
+ commit 67a59023164bc039129b34a742f06ba376687684
279
+ Author: Eric Wong <normalperson@yhbt.net>
280
+ Date: Wed Feb 17 00:37:43 2010 -0800
281
+
282
+ make POSIX_MQ#dup and POSIX_MQ#clone no-op
283
+
284
+ It'll cause problems for the automatic mq_close() during GC
285
+ otherwise, as dup(2) on an mqd_t isn't portable.
286
+
287
+ Of course there's no point in cloning or duping, either, as
288
+ mq_send/mq_receive operations are always atomic at the kernel
289
+ level and only one thread can have a notification registered
290
+ for it.
291
+
292
+ commit 05e577616b74bea99a0e43e07f28823ddda1aaf9
293
+ Author: Eric Wong <normalperson@yhbt.net>
294
+ Date: Sat Feb 13 03:39:17 2010 -0800
295
+
296
+ posix_mq 0.3.1
297
+
298
+ This fixes a misuse of the Ruby API leading to memory leaks in
299
+ cases where message queues are continually opened and closed
300
+ throughout the lifetime of the application.
301
+
302
+ Fortunately applications have little reason to repeatedly open
303
+ and close message queue descriptors: they are
304
+ multi-thread/multi-process-safe in every way imaginable and also
305
+ capable of non-blocking operation.
306
+
307
+ commit 9adbee0ab71bf408db5c3befb43b2bab0d86ebb2
308
+ Author: Eric Wong <normalperson@yhbt.net>
309
+ Date: Sat Feb 13 03:30:44 2010 -0800
310
+
311
+ use GC correctly and avoid memory leaks
312
+
313
+ We still need to explicitly free the pointer we're given, and
314
+ not just close the associated file descriptor. Fortunately most
315
+ people to not spend all day opening/closing message queue
316
+ descriptors so this leak may not be noticeable.
317
+
318
+ commit c07cf2979036b9550566d59d6d4899be98f3e553
319
+ Author: Eric Wong <normalperson@yhbt.net>
320
+ Date: Thu Jan 21 20:44:22 2010 -0800
321
+
322
+ add #shift test with destination buffer
323
+
324
+ commit 1d0bf6f1d39e085948008c2c6d381ed929ac109e
325
+ Author: Eric Wong <normalperson@yhbt.net>
326
+ Date: Sat Jan 9 15:16:40 2010 -0800
327
+
328
+ Rakefile: fix raa_update task
329
+
330
+ commit 2e420820d3b3fb228c810937539f95a618a2c271
331
+ Author: Eric Wong <normalperson@yhbt.net>
332
+ Date: Sat Jan 9 22:52:27 2010 +0000
333
+
334
+ posix_mq 0.3.0
335
+
336
+ This release adds a few new API methods, fixes MRI 1.8.6
337
+ support. We should now have full feature parity with
338
+ underlying POSIX message queue C API.
339
+
340
+ * POSIX_MQ#notify(&block)
341
+ RDoc: http://bogomips.org/ruby_posix_mq/POSIX_MQ.html#M000001
342
+ This is only supported on platforms that implement
343
+ SIGEV_THREAD with mq_notify(3) (tested with glibc + Linux).
344
+ Other platforms will have to continue to rely on signal
345
+ notifications via POSIX#notify=signal, or IO notifications
346
+ in FreeBSD (and Linux).
347
+
348
+ * POSIX_MQ#shift([buffer [,timeout]])
349
+ Shorthand for the common "POSIX_MQ#receive.first"
350
+ when you do not care for priority of the received message.
351
+
352
+ Rev, EventMachine and Reactor support are planned for
353
+ Linux, FreeBSD and possibly any other platforms where POSIX
354
+ message queues are implemented with a file descriptor.
355
+
356
+ commit 2c71257b2b95e737088726ffc963b4e72f1b5455
357
+ Author: Eric Wong <normalperson@yhbt.net>
358
+ Date: Sat Jan 9 22:49:49 2010 +0000
359
+
360
+ MRI 1.8 does not have rb_str_flush
361
+
362
+ It's Rubinius-specific and we use rb_str_resize
363
+ there anyways...
364
+
365
+ commit 531106e51e519458d37bed3721da4eff2f163206
366
+ Author: Eric Wong <normalperson@yhbt.net>
367
+ Date: Fri Jan 8 11:30:49 2010 -0800
368
+
369
+ no point in non-blocking for fd notifications
370
+
371
+ It's not needed since the native thread will retry in the
372
+ unlikely case of EINTR/EAGAIN. And writing one byte to a pipe
373
+ that's guaranteed by POSIX to be at least 512 bytes is highly
374
+ unlikely.
375
+
376
+ It's also bad because F_SETFL takes the big kernel lock under
377
+ Linux (and possibly other systems), and doing it unnecessarily
378
+ is a waste of system cycles.
379
+
380
+ commit d03c76ae11ca6294e05262df747e4d43822ada73
381
+ Author: Eric Wong <normalperson@yhbt.net>
382
+ Date: Fri Jan 8 11:25:06 2010 -0800
383
+
384
+ mode_t is usually unsigned
385
+
386
+ Most used open modes are well under INT_MAX, however
387
+
388
+ commit b3c31cf444e2ca3dae0f6d2370944bfbf3382d88
389
+ Author: Eric Wong <normalperson@yhbt.net>
390
+ Date: Thu Jan 7 01:45:25 2010 -0800
391
+
392
+ add POSIX_MQ#shift helper method
393
+
394
+ This acts like POSIX_MQ#receive but only returns the message
395
+ without the priority.
396
+
397
+ commit 3700db51399e4949ed314ad0545d037b7762064e
398
+ Author: Eric Wong <normalperson@yhbt.net>
399
+ Date: Thu Jan 7 09:28:57 2010 +0000
400
+
401
+ POSIX_MQ#notify only works on GNU/Linux for now
402
+
403
+ SIGEV_THREAD is not easy to implement, so many platforms
404
+ do not implement it.
405
+
406
+ commit 40d61f55ac53e3cd2f229d0b032da03032e3d53d
407
+ Author: Eric Wong <normalperson@yhbt.net>
408
+ Date: Thu Jan 7 00:37:57 2010 -0800
409
+
410
+ POSIX_MQ#notify block execution on message received
411
+
412
+ This is implementation uses both a short-lived POSIX thread and
413
+ a pre-spawned Ruby Thread in a manner that works properly under
414
+ both Ruby 1.8 (green threads) and 1.9 (where Ruby Threads are
415
+ POSIX threads).
416
+
417
+ The short-lived POSIX thread will write a single "\0" byte to
418
+ a pipe the Ruby Thread waits on. This operation is atomic
419
+ on all platforms. Once the Ruby Thread is woken up from the
420
+ pipe, it will execute th block given to it.
421
+
422
+ This dual-thread implementation is inspired by the way glibc
423
+ implements mq_notify(3) + SIGEV_THREAD under Linux where the
424
+ kernel itself cannot directly spawn POSIX threads.
425
+
426
+ commit d8c8fb4155c1feea454abc3ed3f0a4b26e90be68
427
+ Author: Eric Wong <normalperson@yhbt.net>
428
+ Date: Sun Jan 3 05:26:00 2010 +0000
429
+
430
+ fix warnings on platforms where mqd_t != int
431
+
432
+ The POSIX manpages specify the return values of all
433
+ mq_* functions besides mq_open(3) to be "int", not "mqd_t".
434
+
435
+ commit dbe5ed46e07b853e79e44141924a0166016e3e44
436
+ Author: Eric Wong <normalperson@yhbt.net>
437
+ Date: Sat Jan 2 23:19:34 2010 -0800
438
+
439
+ bump GIT-VERSION-GEN
440
+
441
+ Shouldn't affect most people since they should just
442
+ take code from git...
443
+
444
+ commit fd2fcdeee6b44f7854255cb7e01c81db3cd2d99c
445
+ Author: Eric Wong <normalperson@yhbt.net>
446
+ Date: Sun Jan 3 05:46:45 2010 +0000
447
+
448
+ posix_mq 0.2.0
449
+
450
+ This release fixes notification (un)registration and should be
451
+ fully-supported on modern FreeBSD (7.2+) releases.
452
+
453
+ POSIX_MQ#notify=nil correctly unregister notification requests.
454
+ POSIX_MQ#notify=false now provids the no-op SIGEV_NONE
455
+ functionality. Under FreeBSD, using IO.select on POSIX_MQ
456
+ objects is now possible as it has always been under Linux.
457
+
458
+ commit 350bfc831938d84cc2d478f2cf88583863cb64fb
459
+ Author: Eric Wong <normalperson@yhbt.net>
460
+ Date: Sun Jan 3 05:35:50 2010 +0000
461
+
462
+ support POSIX_MQ#to_io under FreeBSD
463
+
464
+ FreeBSD implements an __mq_oshandle(mqd_t mqd) function
465
+ to convert mqd_t to integer file descriptors.
466
+
467
+ commit 26015d39e9c848a536b4ea44802f858a4e6e74f7
468
+ Author: Eric Wong <normalperson@yhbt.net>
469
+ Date: Sun Jan 3 04:17:56 2010 +0000
470
+
471
+ fix build under FreeBSD 7.2
472
+
473
+ FreeBSD seems to need some files explicitly included.
474
+
475
+ commit e7ac70686b3c09953f6c41966dbd77f77368a9ec
476
+ Author: Eric Wong <normalperson@yhbt.net>
477
+ Date: Sat Jan 2 21:19:58 2010 -0800
478
+
479
+ doc: FreeBSD-specific notes + example code
480
+
481
+ commit 6f24b5626f89fbb6445c0943ba71f56c8945e18f
482
+ Author: Eric Wong <normalperson@yhbt.net>
483
+ Date: Sat Jan 2 20:41:39 2010 -0800
484
+
485
+ Fix mq.notify = nil to unregister notifications
486
+
487
+ "mq.notify = false" also works now, doing what
488
+ "mq.notify = nil" used to do (using SIGEV_NONE).
489
+
490
+ I was confused by SIGEV_NONE usage vs using a NULL pointer for
491
+ the notification passed mq_notify(3). SIGEV_NONE does not
492
+ actually unregister, it registers a no-op notification which
493
+ prevents other processes from taking us.
494
+
495
+ This also fixes the test case to pass under both Linux and
496
+ FreeBSD.
497
+
498
+ commit 522d4d1472c216bd95a16ca5b118bc14693aad64
499
+ Author: Eric Wong <normalperson@yhbt.net>
500
+ Date: Sat Jan 2 02:33:23 2010 -0800
501
+
502
+ initial commit
data/GIT-VERSION-FILE CHANGED
@@ -1 +1 @@
1
- GIT_VERSION = 0.5.1
1
+ GIT_VERSION = 0.6.0
data/GIT-VERSION-GEN CHANGED
@@ -1,7 +1,7 @@
1
1
  #!/bin/sh
2
2
 
3
3
  GVF=GIT-VERSION-FILE
4
- DEF_VER=v0.5.1.GIT
4
+ DEF_VER=v0.6.0.GIT
5
5
 
6
6
  LF='
7
7
  '