You cannot select more than 25 topics Topics must start with a letter or number, can include dashes ('-') and can be up to 35 characters long.
koreader/spec/unit/uimanager_spec.lua

420 lines
15 KiB
Lua

describe("UIManager spec", function()
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
local TimeVal, UIManager
local now, wait_until
local noop = function() end
setup(function()
require("commonrequire")
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
TimeVal = require("ui/timeval")
UIManager = require("ui/uimanager")
end)
it("should consume due tasks", function()
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
now = TimeVal:now()
local future = TimeVal:new{ sec = now.sec + 60000, usec = now.usec }
local future2 = TimeVal:new{ sec = future.sec + 5, usec = future.usec}
UIManager:quit()
UIManager._task_queue = {
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
{ time = TimeVal:new{ sec = now.sec - 10, usec = now.usec }, action = noop, args = {}, argc = 0 },
{ time = TimeVal:new{ sec = now.sec, usec = now.usec - 5 }, action = noop, args = {}, argc = 0 },
{ time = now, action = noop, args = {}, argc = 0 },
{ time = future, action = noop, args = {}, argc = 0 },
{ time = future2, action = noop, args = {}, argc = 0 },
}
UIManager:_checkTasks()
assert.are.same(2, #UIManager._task_queue, 2)
assert.are.same(future, UIManager._task_queue[1].time)
assert.are.same(future2, UIManager._task_queue[2].time)
end)
it("should calcualte wait_until properly in checkTasks routine", function()
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
now = TimeVal:now()
local future = TimeVal:new{ sec = now.sec + 60000, usec = now.usec }
UIManager:quit()
UIManager._task_queue = {
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
{ time = TimeVal:new{ sec = now.sec - 10, usec = now.usec }, action = noop, args = {}, argc = 0 },
{ time = TimeVal:new{ sec = now.sec, usec = now.usec - 5 }, action = noop, args = {}, argc = 0 },
{ time = now, action = noop, args = {}, argc = 0 },
{ time = future, action = noop, args = {}, argc = 0 },
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
{ time = TimeVal:new{ sec = future.sec + 5, usec = future.usec }, action = noop, args = {}, argc = 0 },
}
wait_until, now = UIManager:_checkTasks()
assert.are.same(future, wait_until)
end)
it("should return nil wait_until properly in checkTasks routine", function()
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
now = TimeVal:now()
UIManager:quit()
UIManager._task_queue = {
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
{ time = TimeVal:new{ sec = now.sec - 10, usec = now.usec }, action = noop, args = {}, argc = 0 },
{ time = TimeVal:new{ sec = now.sec, usec = now.usec - 5 }, action = noop, args = {}, argc = 0 },
{ time = now, action = noop, args = {}, argc = 0 },
}
wait_until, now = UIManager:_checkTasks()
assert.are.same(nil, wait_until)
end)
it("should insert new task properly in empty task queue", function()
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
now = TimeVal:now()
UIManager:quit()
UIManager._task_queue = {}
assert.are.same(0, #UIManager._task_queue)
UIManager:scheduleIn(50, 'foo')
assert.are.same(1, #UIManager._task_queue)
assert.are.same('foo', UIManager._task_queue[1].action)
end)
it("should insert new task properly in single task queue", function()
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
now = TimeVal:now()
local future = TimeVal:new{ sec = now.sec + 10000, usec = now.usec }
UIManager:quit()
UIManager._task_queue = {
{ time = future, action = '1', args = {}, argc = 0 },
}
assert.are.same(1, #UIManager._task_queue)
UIManager:scheduleIn(150, 'quz')
assert.are.same(2, #UIManager._task_queue)
assert.are.same('quz', UIManager._task_queue[1].action)
UIManager:quit()
UIManager._task_queue = {
{ time = now, action = '1', args = {}, argc = 0 },
}
assert.are.same(1, #UIManager._task_queue)
UIManager:scheduleIn(150, 'foo')
assert.are.same(2, #UIManager._task_queue)
assert.are.same('foo', UIManager._task_queue[2].action)
UIManager:scheduleIn(155, 'bar')
assert.are.same(3, #UIManager._task_queue)
assert.are.same('bar', UIManager._task_queue[3].action)
end)
it("should insert new task in ascendant order", function()
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
now = TimeVal:now()
UIManager:quit()
UIManager._task_queue = {
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
{ time = TimeVal:new{ sec = now.sec - 10, usec = now.usec }, action = '1', args = {}, argc = 0 },
{ time = TimeVal:new{ sec = now.sec, usec = now.usec - 5 }, action = '2', args = {}, argc = 0 },
{ time = now, action = '3', args = {}, argc = 0 },
}
-- insert into the tail slot
UIManager:scheduleIn(10, 'foo')
assert.are.same('foo', UIManager._task_queue[4].action)
-- insert into the second slot
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
UIManager:schedule(TimeVal:new{ sec = now.sec - 5, usec = now.usec }, 'bar')
assert.are.same('bar', UIManager._task_queue[2].action)
-- insert into the head slot
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
UIManager:schedule(TimeVal:new{ sec = now.sec - 15, usec = now.usec }, 'baz')
assert.are.same('baz', UIManager._task_queue[1].action)
-- insert into the last second slot
UIManager:scheduleIn(5, 'qux')
assert.are.same('qux', UIManager._task_queue[6].action)
-- insert into the middle slot
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
UIManager:schedule(TimeVal:new{ sec = now.sec, usec = now.usec - 1 }, 'quux')
assert.are.same('quux', UIManager._task_queue[5].action)
end)
it("should unschedule all the tasks with the same action", function()
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
now = TimeVal:now()
UIManager:quit()
UIManager._task_queue = {
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
{ time = TimeVal:new{ sec = now.sec - 15, usec = now.usec }, action = '3', args = {}, argc = 0 },
{ time = TimeVal:new{ sec = now.sec - 10, usec = now.usec }, action = '1', args = {}, argc = 0 },
{ time = TimeVal:new{ sec = now.sec, usec = now.usec - 6 }, action = '3', args = {}, argc = 0 },
{ time = TimeVal:new{ sec = now.sec, usec = now.usec - 5 }, action = '2', args = {}, argc = 0 },
{ time = now, action = '3', args = {}, argc = 0 },
}
-- insert into the tail slot
UIManager:unschedule('3')
assert.are.same({
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
{ time = TimeVal:new{ sec = now.sec - 10, usec = now.usec }, action = '1', args = {}, argc = 0 },
{ time = TimeVal:new{ sec = now.sec, usec = now.usec - 5 }, action = '2', args = {}, argc = 0 },
}, UIManager._task_queue)
end)
it("should not have race between unschedule and _checkTasks", function()
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
now = TimeVal:now()
local run_count = 0
local task_to_remove = function()
run_count = run_count + 1
end
UIManager:quit()
UIManager._task_queue = {
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
{ time = TimeVal:new{ sec = now.sec, usec = now.usec - 5 }, action = task_to_remove, args = {}, argc = 0 },
{
The great Input/GestureDetector/TimeVal spring cleanup (a.k.a., a saner main loop) (#7415) * ReaderDictionary: Port delay computations to TimeVal * ReaderHighlight: Port delay computations to TimeVal * ReaderView: Port delay computations to TimeVal * Android: Reset gesture detection state on APP_CMD_TERM_WINDOW. This prevents potentially being stuck in bogus gesture states when switching apps. * GestureDetector: * Port delay computations to TimeVal * Fixed delay computations to handle time warps (large and negative deltas). * Simplified timed callback handling to invalidate timers much earlier, preventing accumulating useless timers that no longer have any chance of ever detecting a gesture. * Fixed state clearing to handle the actual effective slots, instead of hard-coding slot 0 & slot 1. * Simplified timed callback handling in general, and added support for a timerfd backend for better performance and accuracy. * The improved timed callback handling allows us to detect and honor (as much as possible) the three possible clock sources usable by Linux evdev events. The only case where synthetic timestamps are used (and that only to handle timed callbacks) is limited to non-timerfd platforms where input events use a clock source that is *NOT* MONOTONIC. AFAICT, that's pretty much... PocketBook, and that's it? * Input: * Use the <linux/input.h> FFI module instead of re-declaring every constant * Fixed (verbose) debug logging of input events to actually translate said constants properly. * Completely reset gesture detection state on suspend. This should prevent bogus gesture detection on resume. * Refactored the waitEvent loop to make it easier to comprehend (hopefully) and much more efficient. Of specific note, it no longer does a crazy select spam every 100µs, instead computing and relying on sane timeouts, as afforded by switching the UI event/input loop to the MONOTONIC time base, and the refactored timed callbacks in GestureDetector. * reMarkable: Stopped enforcing synthetic timestamps on input events, as it should no longer be necessary. * TimeVal: * Refactored and simplified, especially as far as metamethods are concerned (based on <bsd/sys/time.h>). * Added a host of new methods to query the various POSIX clock sources, and made :now default to MONOTONIC. * Removed the debug guard in __sub, as time going backwards can be a perfectly normal occurrence. * New methods: * Clock sources: :realtime, :monotonic, :monotonic_coarse, :realtime_coarse, :boottime * Utility: :tonumber, :tousecs, :tomsecs, :fromnumber, :isPositive, :isZero * UIManager: * Ported event loop & scheduling to TimeVal, and switched to the MONOTONIC time base. This ensures reliable and consistent scheduling, as time is ensured never to go backwards. * Added a :getTime() method, that returns a cached TimeVal:now(), updated at the top of every UI frame. It's used throughout the codebase to cadge a syscall in circumstances where we are guaranteed that a syscall would return a mostly identical value, because very few time has passed. The only code left that does live syscalls does it because it's actually necessary for accuracy, and the only code left that does that in a REALTIME time base is code that *actually* deals with calendar time (e.g., Statistics). * DictQuickLookup: Port delay computations to TimeVal * FootNoteWidget: Port delay computations to TimeVal * HTMLBoxWidget: Port delay computations to TimeVal * Notification: Port delay computations to TimeVal * TextBoxWidget: Port delay computations to TimeVal * AutoSuspend: Port to TimeVal * AutoTurn: * Fix it so that settings are actually honored. * Port to TimeVal * BackgroundRunner: Port to TimeVal * Calibre: Port benchmarking code to TimeVal * BookInfoManager: Removed unnecessary yield in the metadata extraction subprocess now that subprocesses get scheduled properly. * All in all, these changes reduced the CPU cost of a single tap by a factor of ten (!), and got rid of an insane amount of weird poll/wakeup cycles that must have been hell on CPU schedulers and batteries..
3 years ago
time = TimeVal:new{ sec = now.sec - 10, usec = now.usec },
action = function()
run_count = run_count + 1
UIManager:unschedule(task_to_remove)
end,
args = {},
argc = 0
},
{ time = now, action = task_to_remove, args = {}, argc = 0 },
}
UIManager:_checkTasks()
assert.are.same(2, run_count)
end)
it("should clear _task_queue_dirty bit before looping", function()
UIManager:quit()
assert.is.not_true(UIManager._task_queue_dirty)
UIManager:nextTick(function() UIManager:nextTick(noop) end)
UIManager:_checkTasks()
assert.is_true(UIManager._task_queue_dirty)
end)
describe("modal widgets", function()
it("should insert modal widget on top", function()
-- first modal widget
UIManager:show({
x_prefix_test_number = 1,
modal = true,
handleEvent = function()
return true
end
})
-- regular widget, should go under modal widget
UIManager:show({
x_prefix_test_number = 2,
modal = nil,
handleEvent = function()
return true
end
})
assert.equals(2, UIManager._window_stack[1].widget.x_prefix_test_number)
assert.equals(1, UIManager._window_stack[2].widget.x_prefix_test_number)
end)
it("should insert second modal widget on top of first modal widget", function()
UIManager:show({
x_prefix_test_number = 3,
modal = true,
handleEvent = function()
return true
end
})
assert.equals(2, UIManager._window_stack[1].widget.x_prefix_test_number)
assert.equals(1, UIManager._window_stack[2].widget.x_prefix_test_number)
assert.equals(3, UIManager._window_stack[3].widget.x_prefix_test_number)
end)
end)
it("should check active widgets in order", function()
local call_signals = {false, false, false}
UIManager._window_stack = {
{
widget = {
is_always_active = true,
handleEvent = function()
call_signals[1] = true
return true
end
}
},
{
widget = {
is_always_active = true,
handleEvent = function()
call_signals[2] = true
return true
end
}
},
{
widget = {
is_always_active = true,
handleEvent = function()
call_signals[3] = true
return true
end
}
},
{widget = {handleEvent = function()end}},
}
UIManager:sendEvent("foo")
assert.falsy(call_signals[1])
assert.falsy(call_signals[2])
assert.truthy(call_signals[3])
end)
it("should handle stack change when checking for active widgets", function()
-- senario 1: 2nd widget removes the 3rd widget in the stack
local call_signals = {0, 0, 0}
UIManager._window_stack = {
{
widget = {
is_always_active = true,
handleEvent = function()
call_signals[1] = call_signals[1] + 1
end
}
},
{
widget = {
is_always_active = true,
handleEvent = function()
call_signals[2] = call_signals[2] + 1
end
}
},
{
widget = {
is_always_active = true,
handleEvent = function()
call_signals[3] = call_signals[3] + 1
table.remove(UIManager._window_stack, 2)
end
}
},
{widget = {handleEvent = function()end}},
}
UIManager:sendEvent("foo")
assert.is.same(call_signals[1], 1)
assert.is.same(call_signals[2], 0)
assert.is.same(call_signals[3], 1)
-- senario 2: top widget removes itself
call_signals = {0, 0, 0}
UIManager._window_stack = {
{
widget = {
is_always_active = true,
handleEvent = function()
call_signals[1] = call_signals[1] + 1
end
}
},
{
widget = {
is_always_active = true,
handleEvent = function()
call_signals[2] = call_signals[2] + 1
end
}
},
{
widget = {
is_always_active = true,
handleEvent = function()
call_signals[3] = call_signals[3] + 1
table.remove(UIManager._window_stack, 3)
end
}
},
}
UIManager:sendEvent("foo")
assert.is.same(1, call_signals[1])
assert.is.same(1, call_signals[2])
assert.is.same(1, call_signals[3])
end)
it("should handle stack change when broadcasting events", function()
UIManager._window_stack = {
{
widget = {
handleEvent = function()
UIManager._window_stack[1] = nil
end
}
},
}
UIManager:broadcastEvent("foo")
assert.is.same(#UIManager._window_stack, 0)
-- Remember that the stack is processed top to bottom!
-- Test making a hole in the middle of the stack.
UIManager._window_stack = {
{
widget = {
handleEvent = function()
assert.truthy(true)
end
}
},
{
widget = {
handleEvent = function()
assert.falsy(true)
end
}
},
{
widget = {
handleEvent = function()
assert.falsy(true)
end
}
},
{
widget = {
handleEvent = function()
table.remove(UIManager._window_stack, #UIManager._window_stack - 2)
table.remove(UIManager._window_stack, #UIManager._window_stack - 2)
table.remove(UIManager._window_stack, #UIManager._window_stack - 1)
end
}
},
{
widget = {
handleEvent = function()
assert.truthy(true)
end
}
},
}
UIManager:broadcastEvent("foo")
assert.is.same(2, #UIManager._window_stack)
-- Test inserting a new widget in the stack
local new_widget = {
widget = {
handleEvent = function()
assert.truthy(true)
end
}
}
UIManager._window_stack = {
{
widget = {
handleEvent = function()
table.insert(UIManager._window_stack, new_widget)
end
}
},
{
widget = {
handleEvent = function()
assert.truthy(true)
end
}
},
}
UIManager:broadcastEvent("foo")
assert.is.same(3, #UIManager._window_stack)
end)
it("should handle stack change when closing widgets", function()
local widget_1 = {handleEvent = function()end}
local widget_2 = {
handleEvent = function()
UIManager:close(widget_1)
end
}
local widget_3 = {handleEvent = function()end}
UIManager._window_stack = {
{widget = widget_1},
{widget = widget_2},
{widget = widget_3},
}
UIManager:close(widget_2)
assert.is.same(1, #UIManager._window_stack)
assert.is.same(widget_3, UIManager._window_stack[1].widget)
end)
end)