File: RELEASE_NOTES

package info (click to toggle)
cwirc 2.0.0-3
  • links: PTS
  • area: main
  • in suites: squeeze
  • size: 640 kB
  • ctags: 589
  • sloc: ansic: 5,423; makefile: 299
file content (293 lines) | stat: -rw-r--r-- 14,664 bytes parent folder | download | duplicates (5)
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
2.0.0:	The previous version has been used for nearly two years without any
	reported bugs, so it can be considered quite stable. Only a minor
	function was added at users' request in this release, so it is now
	pretty much feature-complete. Hence the jump in major version number.

1.8.8:	This release fixes a rather blatant bug corruption bug that somehow
	managed to surface only now. It would occur when a serie of backlogged
	CWirc frames would arrive from the IRC server all at once, right after
	the receiving client's buffer had underrun, during the fadeout period.
	This bug would cause a bunch of "Receiving cw from ..." lines to be
	dumped in the X-Chat window, but with no CW signal for several seconds,
	and would be triggered by a high buffering value and a lagged IRC
	server.

1.8.7:

1.8.6:	CWirc now filters out Morse frames even when the plugin is disabled.
	That means there's always some CWirc code running all the time in
	X-Chat, whereas until now, when the plugin was disabled, it was truly
	inactive.

1.8.5:	The CWirc Debian package is now maintained by Joop Stakenborg PG4I.

1.8.4:	The CWirc frontend now tries to renice itself when it starts. If it
	can (because the binary was installed suid root), the performances for
	the I/O process are improved, which helps solve the bad sound problem
	on system with low-end sound chips. A less kludgy solution would have
	been to increase the sound buffering, but this would have increased the
	lag between local keying and beeping, which rapidly becomes
	disconcerting for high-speed keyers. Right now, the lag is 30ms and
	that is already quite enough to make a noticeable difference between
	keying a real rig and CWirc.

1.8.3:

1.8.2:	The size of the CWirc frontend binary was reduced dramatically by
	very simply tricks, like re-working and optimizing the XPM icons.
	Also, the code was cleaned up, resulting in some improvenemts in
	memory usage.

1.8.1:	An obscure double-to-int rounding bug has been squashed. So obscure in
	fact that nobody ever hit it, but it was there nonetheless.

1.8.0:	The keyer has been totally revamped. Now it proposes truly correct
	modes, and many options to fine-tune it just right.

1.7.7:	A number of people were unhappy with the Curtis mode-B keyer setting,
	as it isn't like a conventional dit memory, or a Super CMOS mode-B.
	Also, the simple "mode-B" toggle button wasn't a very clear way to set
	the keyer's mode. Therefore, 2 new keyer modes have been added and a
	menu is now used to select them.

1.7.6:

1.7.5:	The sidetone mode and the decoder decoding the local key was something
	many people have wanted for a long time. Now it's possible to see what
	you key, without sending anything on the channel, which is good for
	training to key.

1.7.1:	The addition of the Hellschreiber extension, and the intolerance of the
	Hellschreiber mode for timing errors, has revealed bad inaccuracies in
	the sent and received event values, mainly due to rounding floats into
	ints when sending, and ints back to floats when receiving, with no
	correction for the timing drift this introduces. Timing drifts are now
	handled correctly, making Hellschreiber transmissions almost perfect.

1.7.0:	The frame format has been extended to conserve bandwidth when
	transmitting many keyups/keydowns per frames. The base frame format is
	entirely compatible with the previous formats used so far. The extended
	frame format, however, isn't. Consequently, newer CWirc version will be
	able to receive transmissions from clients up to version 1.6.1, but
	older clients won't be able to receive all the frames from the newer
	versions.

	The new CWirc extension mechanism allows external programs to interact
	with CWirc : they can receive the sound CWirc generates using a ring
	buffer and 2 semaphores to transfer audio data, and to synchronize
	themselves with CWirc, and a flag to key CWirc. Extension programs (or
	synlinks to them) must be installed in the directory specified in the
	Makefile for CWirc to see them and list them in the extensions tab.
	When CWirc starts an extension program, it passes it "--cwirc" followed
	by the shared memory id of the extensions API. Only one extension
	program can run at any given time.

	The extensions API was made specifically to interface gMFSK and CWirc,
	in order to do FeldHell-over-IP : Feldhell is the only digital mode
	that's slow enough and close enough to CW to work over IRC.

1.6.1:	NetBSD has its own sound API that is not OSS, the API used by CWirc.
	However, it has an OSS emulation library (libossaudio) that allows
	Linux programs to be compiled without modification. Therefore, in
	order to build and use CWirc under NetBSD, the "-lossaudio" flag must
	be passed to the linker.

1.6.0:

1.5.0:	FreeBSD doesn't seem to have a native RTC mechanism to provide real-time
	periodic interrupts. The only thing it seems to have is a Linux /dev/rtc
	emulator that only has ioctls to start the periodic interrupts, but none
	to stop them, which makes it unsuitable for CWirc. Therefore, the
	soundcard-less operation for FreeBSD has been disabled.

	Also, FreeBSD has different dsp and tty device names, so these are
	reflected in the default CWirc settings.

	Finally, this release has been tested under FreeBSD 5.1. I am told
	FreeBSD 4.x boxes have slight differences in the naming of devices
	nodes in /dev, but the initial FreeBSD port was done on one of these
	boxes, so I believe it works. I'm a FreeBSD newbie though, so don't
	take my word for it. You might have to hack around to get it to go.

1.4.2:

1.4.1:

1.4.0:

1.3.0:	The new signal strength simulation is based on the physical distance
	between stations. If the signal strength is completely artificial and
	made up, the algorithm to calculate the distance between two stations
	on the planet is accurate.

	The QRM generator generates pseudo pink noise by downsampling white
	noise and upsampling it back. It's not an accurate radio noise but it's
	much less CPU-intensive than "proper" sound generation with FFTs.

1.2.3:	The automatic decoder now has a real "inertia" built-in, i.e. it tends
	to resist changing decoding speed. As a result, transient timing errors
	are more easily ironed out. Also, to decode bad straight keying better,
	it now works on the assumption that beginners follow the "music" even
	when they key dits or dahs way too short, resulting in spaces that are
	too long and would normally trigger the decoding of the current
	character. The decoder now silently allows longer spaces if it sees
	characters that it thinks are too short. This feature is on all the time
	as it doesn't impact the decoding of good keying by definition.

1.2.2:

1.2.1:	CWirc is now more conservative with CPU, so it's possible to use it on
	slower computers, and faster computers with a high load, without sound
	breakup. This was done mainly by replacing trigonometry function calls
	by precalculated tables in tight loops, and by reducing the amount of
	GTK pixmap redraws in the S-meter drawing code.

1.2.0:	This version can now generate sounder clicks as well as regular morse
	beeps. The sounder click samples come from an actual sounder, and the
	timing of its arm going down and up as the coil is energized and
	de-energized is reproduced accurately in CWirc.

	In the process of extending the sound routines, the regular beeping
	sound has also been cleanup up to fade in and out softly, removing
	the annoying clicks at the end of each beep.

1.1.0:	The CW frame protocol has been extended to pass an explicit callsign
	along IRC message. The feature is of little value alone, but it will
	allow single repeater bots to forward message from multiple callsigns
	on non-CWirc morse networks. Note that the explicit callsign is
	scrambled in the IRC message, so casual IRC chatters don't immediately
	see who it is from.

1.0.1:	Just a few minor cosmetic changes in this release.

1.0.0:	This is a major code change. The user interface has been recoded as a
	GTK+ gui, which meant splitting the X-Chat plugin part of CWirc from
	the rest, so as to avoid problems between the X-Chat GTK+ session and
	the CWirc GTK+ one.

	Also, in this new version, the I/O process has been recoded to never
	stop and be reconfigurable on the fly. Therefore, even if it doesn't
	perform because of some error, the gui stays up and can reconfigure it.
	As a result, it's now possible to configure CWirc entirely from the gui.
	(no more editing ~/.cwircrc).

0.9.2:

0.9.1:  In the previous version, if the sounder was activated when the plugin
        was disabled, it would continue beeping. This has been fixed by making
        the I/O process terminate gracefully and turn off the sounder before
        exiting. The main process doesn't abruptly kill the I/O and user
        interface processes anymore, it now asks them to terminate and waits
        for them to die, which is cleaner anyway.

	Also, previous versions used up nearly 2M of memory in one shared memory
	block, due to over-generous buffering and way to many concurrent senders
	which was ridiculous. The shared memory usage has been brought down to
	less than 50K.

0.9.0:	It is now possible to output the received or sent morse code on an
	external sounder, or CW oscillator, throught the RTS line on a serial
	port. It's also possible to do without a soundcard altogether, but in
	this case, CWirc needs to use /dev/rtc to get its timing. Since it uses
	the periodic interrupt at 512Hz, it'll need special privileges set in
	/proc/sys/dev/rtc/max-user-freq. Since multiple inputs and outputs where
	getting confusing, 2 new variables have been added to the configuration
	file (cwinput and cwoutput) to clarify things.

0.8.2:

0.8.1:	The details of exactly what iambic timings are implemented in CWirc are
	now included.

0.8.0:	The new morse frame format uses 1/3rd less bandwidth than the previous
	one. The tests I've run show that one can send a continuous stream of
	dits at 30wpm without being throttled or kicked on the most common
	IRC servers, which means that in a real QSO, sustained speeds of 40/45
	wpm can be expected. That should be more than enough for most operators.

0.7.0:	The sound device is now used at 44.1KHz / 16 bit / stereo, in order to
	increase the sound fragment size to 256 bytes and still maintain a
	timing precision in the order of one millisecond. As a result, since
	44100 Hz isn't 1000 times a power of 2 like 8000 Hz, the entire I/O,
	encoding and decoding sections have been adapted to cope with timing
	durations that aren't exactly one millisecond. This was necessary for
	broken or cheap sound cards that can't cope with very small sound
	fragments. Since it's usually admitted that all sound devices should be
	able to cope at least with 256 byte fragments, that's why CWirc now
	uses the most common sound format that consumes as many bytes per
	seconds as possible.

	The Makefile has been modified to ask the user to set the paths to
	xchat-plugin.h and the base installation directory for X-Chat. I've
	just noticed that the Debian install package I use installs X-Chat in
	/usr, while the X-Chat tarball installs in /usr/local and doesn't copy
	the .h file into the install directory. So the user has to set the
	correct path for his/her machine first, and the Makefile now makes sure
	of that.

0.6.0:	An automatic morse decoder has been added. It decodes from the overall
	signal received, so it's useless in a pile-up, but it works rather
	well when receiving cw from a single sender who keys cleanly and
	regularly. Some people have asked for this, to help them copy fast
	iambic keyers.

0.5.1:	

0.5.0:	I finally modularized the code. The single "cwirc.c" file was quickly
	becoming spaghetti code. That gave me the occasion to review all the
	code, interactions between processes, and clean/simplify/better the
	whole thing. A fake S-meter has been added to the control panel : it's
	basically cosmetic, as there isn't much use for RST reports on IRC
        obviously, but it helps feel in front of a real trx, while the static
        panel doesn't.

0.4.2:

0.4.1:
        There's now a delay between the moment a cw frame is received and the
        moment it starts being played on the sound card. This allows for some
        lag in the connection to the IRC server. Before that, IRC servers had
        to be nearly flawlessly snappy to sustain the morse rate without
        choppiness. It does increase the lag between keying a frame on one end
        and hearing it on the other though.

0.4.0:	CWirc can now be used solely with a mouse, without a real key connected
	to a serial port. Many people have asked for this.

0.3.0:	The bug fixes in 0.2.1 didn't quite do it. After much experimenting, I
	determined that X-Chat just isn't thread-safe. Even spawning a thread
	that does nothing but sleep in a loop makes X-Chat segfault randomly
	after a while. So I turned the threads into normal heavy-weight
	processes and it appears rock-stable now. It's not as efficient in
	terms of memory or CPU, but at least it works well. The frame format
	is now more compact and many IRC servers that would kick the client
	when keying fast don't anymore as a result.

0.2.1:	The cleanup code in the standalone version of CWirc was always followed
	by the termination of the program. In the plugin, it may be called again
	and again. As a result, the code that left variables hanging was
	causing the plugin and X-Chat to segfault regularly when doing /CWENABLE
	several times. This is fixed.

0.2.0:	CWirc is now an X-Chat plugin. It makes much more sense than creating a
	brand new standalone IRC client, since X-Chat is full-featured, has
	a fairly complete plugin interface and allows plugins to do pretty much
	whatever they wants in terms of threads and file accesses.
	
	Corrected the iambic code, so that even a light tap on one paddle while
	the other is actuated ends up sending a single inverted element in the
	morse stream.

0.1.0:	Added support for iambic keyers in mode A or B. Mode-A implemetation is
	straightforward, and I implemented mode-B timing as I believe it should
	correctly be, which is that the extra dit or dah sent after releasing
	squeezed paddles is sent only if the paddles are released after the
	first half of the last element has been sent.

0.0.2:	The code was broken in the first release : users couldn't set the
	serial port the morse key is connected to in the config file. This is
	corrected.

0.0.1:	First release of CWirc. The user interface is minimalistic, the code is
	still very unpolished and the buffering and timing code isn't very
	robust. The basics work though.