1 2008 November 14 2.2rc4
2 Change the way that we sleep/wait for activity, to use select
3 rather than a fixed-length sleep, supplied by Julien Blache of
8 The attached patch for wmacpi makes it use select() instead of
9 sleeping. It also adjusts the timeout of the select() call
10 depending on the blink & scroll options.
12 If scrolling is not activated, there's no reason to wake up
13 10x/sec. If blinking is activated, waking up once per second is
14 enough. If blinking is not activated, then we can wake up just to
15 update the data and we're fine.
17 With scrolling and blinking disabled, this saves a number of
18 wakeups and helps battery life. Unfortunately there's another
19 source of wakeups that's probably due to libdockapp that still
20 makes 6 wakeups/seconds (in another thread it seems).
22 With this patch wmacpi makes a best effort to match the sample
23 rate set by the user, but that's hardly an issue I think.
25 I haven't released this patch yet in Debian, so if you like it and
26 want to roll out a release, feel free :) Otherwise I'll add it to
35 Some fixes for the sysfs interface support, supplied by
38 > Okay, I finally found time to make a new release including this
39 > patch - 2.2rc2. It's up on my website now.
41 I don't remember sending you the updated patch, as it turned out that
42 you can have different attributes in sysfs depending on what the
45 The current version is attached.
53 Support for the sysfs interface that became mandatory with kernel
54 2.6.24 (patch supplied by jblache@debian.org).
56 A number of graphics fixes also from jblache@debian.org.
58 Removed support for hardware reported critical battery status.
61 Major changes to command line handling and to the way we use
62 libdockapp, courtesy of Patrice Dumas. This should hopefully fix
63 several long-standing bugs with non-wmaker window managers (in
64 particular Debian bugs #280495 and 293546).
66 Updated to support newer kernels that no longer support
67 /proc/acpi/info - thanks to Samuel Ortiz.
69 Fixed a longstanding issue with redrawing - thanks to Vito
72 This version changes a number of command line options (little used
73 ones, but they're still incompatible changes) due to the
74 libdockapp work. In particular, -w becomes -x, -v becomes -V (and
75 takes an argument rather than being repeated), and -V becomes -v
76 (thanks to libdockapp grabbing -h, -v and -w for itself). Also,
77 long options are now supported for everything in wmacpi (but not
78 acpi, since it doesn't use libdockapp for command line parsing).
81 Added a patch from Patrice Dumas to support newer versions of
84 Updated email address in AUTHORS and manpages.
87 Finalised libdockapp-0.5.0 port - no changes from 2.1rc1.
90 Typo fix in the manpage.
92 Ported to libdockapp-0.5.0 - all this needed was changing the
93 type of the dockapp width and height to unsigned short.
96 Upped version number to 2.0.
100 wmacpi 2.0 goes gold . . .
102 2004 September 28 2.0rc1
103 Added support for switching to capacity mode automatically, on
104 detecting dodgy reports from the battery.
106 Added support for capacity mode on charging, with automatic
107 enabling as per discharging.
111 Hopefully last release before the final 2.0.
113 2004 August 18 1.99r7
114 Implemented the libdockapp port - this seems to close Debian bug
115 #227819, but it hasn't received sufficient testing.
117 Implemented a capacity mode for calculating time remaining (as
118 opposed to the normal current rate mode) - this mode samples the
119 remaining capacity of the battery and notes the time at which it
120 was sampled, and uses a history of samples to estimate the rate of
121 drain on the battery. From there it calculates a value for time
124 Also, various cleanups have gone in:
126 * a reworking of the scrolling code
128 * generic battery number support (rather than just
129 using two pixmaps, one for b1 and one for b2)
131 * stopped the battery glyph from blinking when running on AC
134 Fix for Debian bug #250792 - the parser for the acpi files was
135 too stupid to deal with error conditions. I've now added some
136 basic checking to make it handle the error that caused the bug
137 report, but I'll need to add more later.
140 Collected fixes for the collected fixes below . . .
142 * Manpage fixes, to reflect the reality of the current code.
144 * Code cleanups, to make a few things more sensible. Most notably,
145 the interface for setting the samplerate has changed so that it's
146 no longer inverted and illogical - you now say how many times you
147 want to sample per minute.
149 * Fixed an issue with initialisation - I'd moved the power_init()
150 call below the options parsing code, without dealing with the -m
151 option properly. The end result was that if you told it to monitor
152 a battery number, it would fail saying the battery didn't exist. I
153 moved the check for this out of the options parsing and after the
156 * Fixed a leaking file descriptor in init_ac_adapters.
158 * Implemented a way to handle changing batteries - reinitialise
159 the battery info periodically. I don't know of a better way to do
160 that, since we'd have to do all that parsing anyway to find out if
163 libdockapp is waiting, but I think that's the only change left
164 without more bug repots . . .
167 Collected fixes for various small issues.
169 * Fixed a problem with placement of the window when using the
170 click to place mode - turned out to be a sizehints problem.
172 * Some fixes to the manpage.
174 * Reenabled the CLI build by default - the Debian patches can
175 handle disabling it themselves.
177 * Added a way to disable the scrolling message, since some users
180 I've left the big changes that are needed (like using libdockapp,
181 in the hope that it'll solve the docking problems) until later, so
182 that I can get these smaller fixes out. Hopefully soon . . .
184 2004 January 12 1.99r3
185 . . . and a fix for a fix that didn't fix it . . .
187 2004 January 12 1.99r2
188 Some cleanups and bugfixes found by the wonderful Debian users.
190 It's finally in testing!
192 2003 November 23 1.99
193 Finally claimed the wmacpi name as my own . . .
195 Renamed wmacpi-ng and acpi-ng, renamed the header files, fixed up
198 For the Debian package, also made compilation of the command line
199 tool optional, defaulting to not building it. This is because
200 after the renaming, my acpi clashes with the acpi package that's
201 already in Debian. The command line functionality is now
202 accessible via the -w option to wmacpi.
204 This is wmacpi 1.99, so that I can have a release packaged and in
205 Debian before going to 2.0, so that any bugs that are left can be
206 found by all the extra users.
208 2003 September 26 0.99
209 Fix the last of the old wmacpi code oddities (specifically, the
210 APMInfo struct, which was a completely inappropriate name given we
211 no longer deal with APM, and because of the various code
214 I think this is probably as good as I can make it without getting
215 more feedback and the like, so I'm going to make this version
216 0.99, preliminary to either wmacpi-ng 1.0 or wmacpi 2.0 (depending
217 which direction I go with that).
219 2003 September 23 0.92
220 Fix a few more bugs, and include an option to allow the user to
221 specify how often to sample the ACPI data - some BIOSes disable
222 all interrupts while reading from the battery, which apparently
223 causes some interactivity issues. I have no idea why reading once
224 every three seconds (which is our default), but there've been some
227 Also fixed acpi-ng to properly handle the -a option.
230 Fix a problem I'm seeing with docking the app - if the window name
231 is set to "acpi" rather than "apm" it doesn't dock . . .
234 Make the time display show '--:--' instead of '00:00' when the
235 time remaining is 0 - I think this is reasonable, since it'll only
236 get into this state when the present rate value is 0. This only
237 happens when the battery is fully charged and neither discharging
238 or charging, or when the battery is completely drained. In any of
239 these states the time remaining is of very little interest, so we
240 don't lose anything. We also get to handle the (sadly, very
241 common) case where the ACPI subsystem doesn't report sane values
242 for the things we depend on.
245 Make the time display show nothing (as opposed to 00:00) when the
246 time remaining is unknown, as requested by Emma Jane Hogbin.
248 Respect the critical level specified on the command line, and add
249 a new message to differentiate between the command line critical
250 level and the critical state reported by the battery.
252 Speed up the message scrolling, so that there isn't quite such a
253 painful delay between repeats. Also, up the speed when the battery
254 is low, more when it's critical, and scroll continuously when it's
257 Finally, add support for disabling the blinking power and battery
258 glyphs from the command line, as requested by Sebastian
259 Henschel. We still blink the battery glyph when the battery
260 reports hardware critical level - I think that's worth being a bit
263 Also, added an acpi-ng manpage.
266 Properly fix the AC adapter issue - it's not much use adding the
267 infrastructure and then forgetting to fix the code that uses it.
269 Fix the time disiplay so that it doesn't try to display values
270 greater than 99:59, since the display area won't fit anything
274 Bugfixes, to handle two problems: the case where something like
275 the present rate or some such is "unknown" (reported initially by
276 Emma Jane Hogbin, and where the AC adapter is called something
277 other than "AC" (reported by Sebastian Henschel). This an 'a'
278 release because I can't test these myself . . .
281 Finally got rid of that annoying button - that space now contains
282 a 'B 1' or 'B 2' (only those two at present, since I'm too lazy to
283 fix it so that the number is generic. It should work fine for
284 anyone who doesn't have a freakish system with more than two
287 A few more code cleanups.
290 libacpi cleanups and reworking - we now handle the charging
291 battery case properly, it seems.
293 Also, some attempts to make error printing a bit cleaner; still a
294 long way to go on that, though . . .
297 Removed process_plugin_timer(), since it was doing nothing useful
301 Removed libapm.c - wmacpi-ng is specifically for ACPI, with no APM
305 Some more code cleanups, designed to move libacpi to more of a
306 library than something built into wmacpi-ng. This is useful with
307 the seperate programs, though at present it's not used much.
310 Added a command line tool, acpi-ng to query battery status from
314 Implemented multiple battery support, and averaging of the samples
315 in an attempt to even out the jumpiness of the timer.
318 More major code cleanups: in particular the handling of the power
319 panel and the message has been cleaned up so that it's actually
322 Next step from here is to actually implement handling of multiple
323 batteries, so that if there's a fully charge second battery
324 available it reports the correct time remaining (based on the
325 present rate of power consumption and the sum of the two battery's
326 remaining capacities).
329 Some major reworking of the internals, to help fix the handling of
330 multiple batteries and such things.
333 Code cleanups to fix various problems with corner cases.
336 Added support for multiple batteries: I took the simple route of
337 displaying only one battery per instance - to do more I'd need to
338 hack with the display code, and I'm not ready for that yet.
340 Added a -m option to specify the battery number to monitor.
343 Changed package completely, to support the new ACPI code as of
346 Since this code hasn't been touched in more than a year, I figure
347 I might as well have a go at hacking on it . . .
349 -- Simon Fowler, <simon@dreamcraft.com.au>
352 Updated ACPI statistics gathering code for the latest ACPI patch from
353 Intel. Now uses/checks for subsystem version 20020214+. Redone the version
357 Little fix to ACPI version check so that I don't have to make a new
358 version each time a new kernel comes out
361 Split ACPI and APM portions in different source files
362 Added power_init() function - handles checking subsystem presence,
363 version, etc. Enumerates batteries in ACPI implementation
364 ACPI: Beginnings of multiple battery support - not fully implemented
365 ACPI: Get the battery ID from /proc/acpi/battery instead of hard-coding
366 it in - also helps with multi battery stuff
367 Added some stupid command line help
368 Added -c option to set critical battery alarm level. Defaults to
369 10 percent, and added -b option to make noise when this critical level
371 APM: Updated for critical alarm stuff
372 Made a 20 color "low" quality pixmap for machines running 8bit displays
373 Updated README with some stuff, made separate INSTALL file
376 Update for intel ACPI release version 20010313
377 Minor updates for new /proc layout
380 Update for intel ACPI release verion 20010302
381 Proc format changed a little bit, so updates were necessary.
382 Added a check for specific acpi subsystem version - since each
383 new version seems to have a new interface, better be safe than
385 As usual, please test and report to me.
387 This ONLY works with linux acpica 20010302. Download this version from
388 http://developer.intel.com/technology/iapc/acpi/downloads.htm - it's a
392 Finally read through the ACPI specification. Hah, they had all my
393 questions answered :) Thus,
395 Confirmed remaining percentage formula (correct)
397 Now calculate "time remaining" according to the spec formula
399 Removed "battery load" hack since we now have "time remaining"
401 Fixed bug in display_percentage (10% didn't display right)
403 Released version 1.21
406 Added support for ACPI. This is very untested, and not very clean.
407 This has been tested on machines with only ONE battery, and only
408 with kernel 2.4.2 (that's where ACPI started to work for me)
410 Added some hackery to do with ACPI. Check README.
415 Fixed all the stuff so that wmapm works as advertised:
416 Fixed dumb bug where after charging was done new message didn't scroll.
418 Added support for fucked APM bioses that display stuff wrong. (See
419 README to learn how to enable it. Obviously, it's disabled by default)
421 Reorganized apminfo structure and got rid of redundancy.
423 Reorganized acquire_apm_info. All APM data is processed there.
425 Fixed display_state to do the Right Thing.
427 Fixed red blinking button to actually blink on critical battery, and
428 to stop when it isn't critical anymore. I don't think it ever worked