Add validation to control objects #9

ywwg · 2013-06-08T22:17:05Z

No description provided.

rryan · 2013-06-08T22:20:23Z

mixxx/src/control/control.cpp

@@ -93,6 +96,7 @@ void ControlDoublePrivate::set(const double& value, QObject* pSender) {
    if (pBehavior && !pBehavior->setFilter(&dValue)) {
        return;
    }
+    if (m_pValidator && !m_pValidator->validateChange(dValue)) return;


This has a race condition -- m_pValidator could become NULL between the m_pValidator check and m_pValidator->validateChange call. You should copy it to a local first.

Also not a giant fan of the one line if-statement/body ;).

that's approved google style!

+1 for always use braces
if ( ... ) {
return;
}

Good ..

for recognising control flow.

clean diffs if we insert a statement before return later

set() should become bool set() to inform the caller if the set was successful

Yea :) it's a personal preference and not in the Google C++ style guide. My eyes can easily scan over a 1-liner with return dangling at the end while a block is harder to miss.

daschuer · 2013-06-09T10:21:12Z

I wander how much flexibility of the control object we really need (vs. simplicity and performance).
This affects the behavior and the validator. I think in general if it is good to have them exactly there but with this NULL pointer approach we have to deal with updates for the objects when they are not fully setup.

I am not sure if this is possible, but I would really like to have an owner/user concept for the control concept.
The owner is responsible to setup the control object and once it is finished, it can be released for usage.

What is the difference between validator and behaviour? Is it possible to join them?

rryan · 2013-06-11T02:37:24Z

I think validator and behavior are useful to keep separate. I think of validators as a user-defined validation function while the behavior is more intrinsic to the type of the control.

ywwg · 2013-06-11T04:10:25Z

Updated with some notes addressed. Still no enable/disable, but with better pointer ownership and cleaner implementation.

rryan · 2013-06-11T04:21:19Z

mixxx/src/controlobject.h

@@ -26,6 +26,8 @@
 #include "controllers/midi/midimessage.h"
 #include "control/control.h"

+class ControlValidator;


Since you are providing an inline definition of a method that uses the class you can't forward declare it. The only reason this works is this file includes control.h which includes the controlvalidator.h.

rryan · 2013-06-11T04:25:12Z

Nice! But this still has the lifecycle and threading issues I pointed out in the last round of comments. The validator outlives the EngineBuffer so any sets to the control after the EngineBuffer is deleted will segfault. Also, no thread should touch EngineBuffer except the callback thread and letting the valdiator have a pointer to EngineBuffer means any thread that changes the play control can call code in EngineBuffer.

I think the right solution here is either an enabled/disabled property for controls or an EnabledValidator / DisabledValidator that you set on the play control when the state changes within EngineBuffer.

ywwg · 2013-06-11T12:25:59Z

OK, I thought adding the check in the validator for if(m_pEngine ==NULL) was enough to solve the first problem.

but you're right, a more generic enabled/disabled is needed.

What about controls that require more validation than yes, no, however, like the seek validation?

ywwg · 2013-06-11T13:20:11Z

mixxx/src/control/controlvalidator.h

+
+class ControlValidator {
+  public:
+    ControlValidator() { }


that's weird because when I didn't have the constructor/destructor the compiler complained about possible undefined behavior without a virtual destructor

Yea you're right, it should have a virtual destructor. I've done that elsewhere in Mixxx and I should fix that too.

daschuer · 2013-06-11T13:58:46Z

Hi Owen,

I had time to think a little bit about your validation concept ... and I think it will not work in all cases.
Sorry about this late concerns.

We have always the tiny moment between validation and the actual set of the new value. The high priority audio thread can always interrupt the GUI thread in particular this moment. So all validation might be invalid after that.

The problem in the current trunk solution is that parts of mixxx rely on the play control as indicator for play. e.g. The he feedback LED might be on without a playing track.

One solution might be to handle controls like play as the user command. (You can't stop the user to press play) In addition we need a "playing" or "play_fb"control, which has always the value for the play button feedback LED.

Since Mixxx can only react on play commands each time the audio callback is executed. It is fine to evaluate the "play" control there and set the feedback "playing" and the "play" control back to the real "Play value from the engine buffer.

Kind regards,

Daniel

rryan · 2013-06-11T14:39:40Z

The fact that validation can be interrupted after it runs is fine I think. If the validator validates a value, it has validated that value regardless of whether the control has been updated in the meantime. Take for example playposition. The user wants to seek to 0 and the current value is 0.5. Let's say there's a validator that makes sure that you set it to a positive number (contrived I know since we support the lead-in) but it validates 0 then is interrupted. The engine runs and updates playposition to 0.6. The GUI thread resumes, and the set to 0 is completed. It is still valid to seek to 0 because that's what the user requested.

rryan · 2013-06-11T14:47:30Z

Making a play indicator control would work but there is already tons of existing state (skins, controller scripts) using play as a toggle button and indicator combined.

daschuer · 2013-06-11T14:55:07Z

Hi RJ,

in your case everything is fine, because we have several seek controls for the user command and the play position control for the "truth" feedback. If a user seeks a track to 0.5 which is currently unloaded nothing bad will happen.

Owens Validator concept is fine if the validate rules are not changed during run time.

Introducing a play indicator will not effect the current solutions in skins if they are not changed. The still have the "problem" that the playbutton is lit when no track is loaded for a short moment, but so what. All the critical things like AutoDJ and also new skins could work with the new "play indicator"

ywwg · 2013-06-11T15:26:25Z

OK I've added enabled/disabled. I'm worried that we might want to call set() and overwrite the value even if the control is disabled. See how I had to be careful where to set enabled/disabled on the play button.

daschuer · 2013-06-11T19:16:54Z

As I pointed out, there might be a problem with interrupting after changing m_Enabled.
If we offer a general solution for this, it opens the door for race conditions later on other controls as well.

What are the controls which have a benefit of this? Where it is really needed?

We may offer a thread save solution for this: We may put checking m_Enabled and writing in a critical section.
But I like a clean "play indicator" solution more.

ywwg · 2013-06-11T19:56:21Z

I would like to have enable/disable for:
Places I think we should use enable/disable:
play, cue, ff, rew, ie all playback controls (track not loaded)
all EQ controls (eq bypass)
vinyl controls (vinyl not active)
talk button (no mic connected)

daschuer · 2013-06-12T07:32:51Z

If we take a look at rotary switches. There is a problem with the "Enabled solution", because if a control is disabled the requested value is lost. But the rotary on the midi controller is still pointing to the requested value.
If the CO is enabled again, there is no way to set it to the requested value. The user has turn the button again to receive midi updates.

I have made some drawings with the free yEd, shared with you at
https://docs.google.com/a/mixxx.org/file/d/0B487gWGL6DsXV2FJU2tfbnhkOVU/edit?usp=sharing
https://docs.google.com/a/mixxx.org/file/d/0B487gWGL6DsXeWNMUVVNZW84N0E/edit?usp=sharing

If we finally fix the race condition, all solutions would work.
Here some comments:

Enabled solution.
** It is hard to make it thread save and non blocking
** Fully compatible the current scripts
** It is not clear what Play = 1.0 is, because Engine Buffer may playing or a play command may requested
Feedback solution
** No race condition
** Fully compatible the current scripts
** Can distinguish between play request and playing
** Overhead due to a additional CO
Tree State Solution
** No race condition
** Native atomic writes on 32 bit CPUs
** Not compatible the current scripts (Adapter Required)
** Can distinguish between play request and playing
** New API: Request() / Confirm() / GetCurrent() / GetRequest()
** Requested value is stored

I am not sure whats my favorite, but I am heading towards the three state solution, because it represents a Toggle button with a Feedback LED the best.

ywwg · 2013-06-12T13:55:32Z

Well I guess for EQ it's more complicated because it's fine to change the value via midi while it's disabled, but it shouldn't be possible to change it through the GUI. But for other controls if the user disables the control and then twists a knob, they shouldn't be surprised that when they reenable the control the value isn't correct. That's what midi-takeover is for.

as for the multithreading stuff, I don't really understand the issues so I'll let you guys figure it out. If it was up to me I would just have a system where we still have mutexes and try to move away from locks in the engine gradually. trying to predict all of these race conditions just gives me a headache.

daschuer · 2013-07-28T19:53:31Z

Has any one read:
https://docs.google.com/a/mixxx.org/file/d/0B487gWGL6DsXeWNMUVVNZW84N0E/edit?usp=sharing

I have currently exactly this problem with the play button in the https://github.com/daschuer/mixxx/tree/advanced_autodj_2 branch. I need the play state of each deck, not if there is a pending play request.

After new thoughts about it, I think while "Tree State Solution" is closes to the truth, it brakes the most with the current view of a CO -> a single global" So I like currently the feedback sloution most. because it is straight forward and easy to understand.

I will prepare a branch for review, that introduce a new "play_fb" control and leave the "play" control unchanged.

daschuer · 2013-08-20T22:00:40Z

can we close this one?

ywwg · 2013-08-21T16:59:17Z

oh yeah

There is no need for signal upstream shoutcast because user can't change options when shoutcast sending is on-air.

Some beat size box improvements

Use QAtomicInt

ColumnCache: Verify consistency of all column maps

…h sync When loading a track that is not yet present in the library (and thus doesn't have any BPM because it hasn't been analyzed yet) while another deck is playing and both decks have sync enabled, a debug assertion is triggered: DEBUG ASSERT: "isValid()" in function double mixxx::Bpm::value() const at src/track/bpm.h:53 Aborted (core dumped) The backtrace looks as follows: #0 0x00007f175c87234c in __pthread_kill_implementation () at /usr/lib/libc.so.6 #1 0x00007f175c8254b8 in raise () at /usr/lib/libc.so.6 #2 0x00007f175c80f534 in abort () at /usr/lib/libc.so.6 #3 0x00007f175cf05ee4 in qt_assert(char const*, char const*, int) () at /usr/lib/libQt5Core.so.5 #4 0x000055deb2e67e1c in mixxx::(anonymous namespace)::handleMessage(QtMsgType, QMessageLogContext const&, QString const&) (type=<optimized out>, context=<optimized out>, input=<optimized out>) at src/util/logging.cpp:355 #5 0x00007f175cf47128 in () at /usr/lib/libQt5Core.so.5 #6 0x00007f175cf3fd8a in () at /usr/lib/libQt5Core.so.5 #7 0x00007f175cf06526 in QMessageLogger::critical(char const*, ...) const () at /usr/lib/libQt5Core.so.5 #8 0x000055deb2e5c720 in mixxx_debug_assert(char const*, char const*, int, char const*) (assertion=assertion@entry=0x55deb39bd0db "isValid()", file=file@entry=0x55deb39bbf30 "src/track/bpm.h", line=line@entry=53, function=function@entry=0x55deb39bbf08 "double mixxx::Bpm::value() const") at gsrc/util/assert.h:9 #9 0x000055deb2ee7e7e in mixxx_debug_assert_return_true(char const*, char const*, int, char const*) (function=0x55deb39bbf08 "double mixxx::Bpm::value() const", line=53, file=0x55deb39bbf30 "src/track/bpm.h", assertion=0x55deb39bd0db "isValid()") at gsrc/util/assert.h:18 #10 mixxx::Bpm::value() const (this=<synthetic pointer>) at src/track/bpm.h:53 #11 mixxx::operator*(mixxx::Bpm, double) (multiple=1, bpm=...) at src/track/bpm.h:160 #12 SyncControl::setLocalBpm(mixxx::Bpm) (this=<optimized out>, localBpm=...) at src/engine/sync/synccontrol.cpp:567 #13 0x000055deb34c7ba3 in EngineBuffer::postProcess(int) (this=0x55deb56eb060, iBufferSize=2048) at src/engine/enginebuffer.cpp:1318 #14 0x000055deb3139023 in EngineMaster::processChannels(int) (this=0x55deb5449440, iBufferSize=<optimized out>) at src/engine/enginemaster.cpp:383 #15 0x000055deb31394f7 in EngineMaster::process(int) (this=0x55deb5449440, iBufferSize=iBufferSize@entry=2048) at src/engine/enginemaster.cpp:410 #16 0x000055deb2f91d0b in SoundManager::onDeviceOutputCallback(long) (this=<optimized out>, iFramesPerBuffer=iFramesPerBuffer@entry=1024) at src/soundio/soundmanager.cpp:596 #17 0x000055deb32dd794 in SoundDevicePortAudio::callbackProcessClkRef(long, float*, float const*, PaStreamCallbackTimeInfo const*, unsigned long) (this=0x55deb553e6b0, framesPerBuffer=1024, out=<optimized out>, in=<optimized out>, timeInfo=<optimized out>, statusFlags=<optimized out>) at src/soundio/sounddeviceportaudio.cpp:965 This happens because `newLocalBpm` is invalid when `localBpm` is invalid. Trying to do sync decks while no tempo information is available does not make sense, so we only synchronize decks if the local BPM is available.

rryan reviewed Jun 8, 2013
View reviewed changes

ywwg added 5 commits June 10, 2013 21:55

Implement optional validation of control object changes.

8940f51

Different implementation of validators

a5d9068

make sure to set validator in CDP when it's set in CO

0c4b8f4

start to address notes

5797282

add controlvalidator.h

5822490

address notes

57d9d0a

rryan reviewed Jun 11, 2013
View reviewed changes

ywwg reviewed Jun 11, 2013
View reviewed changes

Add Enabled/Disabled state to ControlPrivateDouble

67cd7d3

daschuer mentioned this pull request Aug 6, 2013

Co request #57

Merged

ywwg closed this Aug 21, 2013

ywwg mentioned this pull request Dec 20, 2013

Refactor the engine to process samples in the range of [-1.0, 1.0]. Fixes Bug #1204039. #126

Merged

uklotzde mentioned this pull request Jul 12, 2014

Update Vestax VCI-300 mapping for 1.12 #280

Merged

daschuer mentioned this pull request Dec 28, 2014

Spinny updates #440

Merged

daschuer mentioned this pull request Jan 8, 2015

New SoundSource/AudioSource API #411

Merged

uklotzde mentioned this pull request Jul 21, 2015

CachingReader: Reduce log spam for corrupt files #652

Merged

daschuer referenced this pull request in daschuer/mixxx Oct 19, 2015

Merge pull request #9 from daschuer/1.12-shoutcast-fixes

8cc4bb3

There is no need for signal upstream shoutcast because user can't change options when shoutcast sending is on-air.

daschuer pushed a commit that referenced this pull request May 25, 2017

Merge pull request #9 from daschuer/beatloop_size

a17d590

Some beat size box improvements

daschuer mentioned this pull request Aug 13, 2017

Multi-broadcasting #1300

Merged

19 tasks

daschuer mentioned this pull request Jun 3, 2018

[WIP] Livemetadata PR #1675

Closed

daschuer mentioned this pull request Jun 11, 2018

Library Layout Redesign Single pane #1117

Closed

11 tasks

Be-ing pushed a commit that referenced this pull request Dec 30, 2018

Merge pull request #9 from daschuer/lp1737537_analysis_revamp

7bf1157

Use QAtomicInt

uklotzde mentioned this pull request Jan 16, 2020

Replace deprecated QSignalMapper with lambda expressions #2427

Merged

uklotzde pushed a commit that referenced this pull request Dec 17, 2020

Merge pull request #9 from uklotzde/column-cache-fix

7a24153

ColumnCache: Verify consistency of all column maps

ronso0 mentioned this pull request Sep 4, 2022

Effects: allow clearing chains with empty '---' preset #10859

Merged

ronso0 mentioned this pull request Sep 13, 2022

adding WVuMeterGL as alternative for WVuMeter #10893

Merged

ronso0 mentioned this pull request May 9, 2023

qopenglwindow-based replacement of qglwidget, full glsl shader based waveforms, vumeters and spinnies #10989

Merged

Swiftb0y mentioned this pull request Aug 7, 2023

GNOME 44 (Wayland): Context menu on tree items in side bar closes immediately #11576

Closed

Frotus mentioned this pull request Jan 27, 2024

Mixxx hang (infinite) at startup, when a secondary monitor is the application window target and RunAs Admin is NOT used #12655

Open

acolombier mentioned this pull request May 22, 2024

Simplify waveform preferences panel #13226

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add validation to control objects #9

Add validation to control objects #9

ywwg commented Jun 8, 2013

rryan Jun 8, 2013

rryan Jun 8, 2013

ywwg Jun 9, 2013

daschuer Jun 9, 2013

daschuer Jun 9, 2013

rryan Jun 11, 2013

daschuer commented Jun 9, 2013

rryan commented Jun 11, 2013

ywwg commented Jun 11, 2013

rryan Jun 11, 2013

rryan commented Jun 11, 2013

ywwg commented Jun 11, 2013

ywwg Jun 11, 2013

rryan Jun 11, 2013

daschuer commented Jun 11, 2013

rryan commented Jun 11, 2013

rryan commented Jun 11, 2013

daschuer commented Jun 11, 2013

ywwg commented Jun 11, 2013

daschuer commented Jun 11, 2013

ywwg commented Jun 11, 2013

daschuer commented Jun 12, 2013

ywwg commented Jun 12, 2013

daschuer commented Jul 28, 2013

daschuer commented Aug 20, 2013

ywwg commented Aug 21, 2013

Add validation to control objects #9

Add validation to control objects #9

Conversation

ywwg commented Jun 8, 2013

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

daschuer commented Jun 9, 2013

rryan commented Jun 11, 2013

ywwg commented Jun 11, 2013

Choose a reason for hiding this comment

rryan commented Jun 11, 2013

ywwg commented Jun 11, 2013

Choose a reason for hiding this comment

Choose a reason for hiding this comment

daschuer commented Jun 11, 2013

rryan commented Jun 11, 2013

rryan commented Jun 11, 2013

daschuer commented Jun 11, 2013

ywwg commented Jun 11, 2013

daschuer commented Jun 11, 2013

ywwg commented Jun 11, 2013

daschuer commented Jun 12, 2013

ywwg commented Jun 12, 2013

daschuer commented Jul 28, 2013

daschuer commented Aug 20, 2013

ywwg commented Aug 21, 2013