Several of the example programs connect the valueChanged signal of a QScrollBar to the display slot, so the LCD number continuously shows the value of the scroll bar. Note that display is overloaded; Qt will select the appropriate version when you connect a signal to the slot. QPushButton ('Manually connected', self) self. AddWidget (self. Manbtn) # Manually connect the clicked signal to it's handler, as we have done # in other examples self. Manbtn, QtCore. SIGNAL ('clicked'), self. Manualslot) # Create a button which we will automatially connect to a slot self. Brokenbtn = QtGui. QWidget has a public slot called show.You can connect your button's clicked signal to your widget's show slot. Read more about signals and slots here. There are two drawbacks in this simple implementation: It's not threadsafe and you cannot disconnect a slot from a signal from within the slot callback. Both problems are easy to solve but would make this example more complex. Using this Signal class other patterns can be implemented easily.
This is the sequel of my previous article explaining the implementation details of the signals and slots.In the Part 1, we have seenthe general principle and how it works with the old syntax.In this blog post, we will see the implementation details behind thenew function pointerbased syntax in Qt5.
New Syntax in Qt5
The new syntax looks like this:
Why the new syntax?
I already explained the advantages of the new syntax in adedicated blog entry.To summarize, the new syntax allows compile-time checking of the signals and slots. It also allowsautomatic conversion of the arguments if they do not have the same types.As a bonus, it enables the support for lambda expressions.
New overloads
There was only a few changes required to make that possible.
The main idea is to have new overloads to QObject::connect which take the pointersto functions as arguments instead of char*
There are three new static overloads of QObject::connect: (not actual code)
The first one is the one that is much closer to the old syntax: you connect a signal from the senderto a slot in a receiver object.The two other overloads are connecting a signal to a static function or a functor object withouta receiver.
They are very similar and we will only analyze the first one in this article.
Pointer to Member Functions
Before continuing my explanation, I would like to open a parenthesis totalk a bit about pointers to member functions.
Here is a simple sample code that declares a pointer to member function and calls it.
Pointers to member and pointers to member functions are usually part of the subset of C++ that is not much used and thus lesser known.
The good news is that you still do not really need to know much about them to use Qt and its new syntax. All you need to remember is to put the & before the name of the signal in your connect call. But you will not need to cope with the ::*, .* or ->* cryptic operators.
These cryptic operators allow you to declare a pointer to a member or access it.The type of such pointers includes the return type, the class which owns the member, the types of each argumentand the const-ness of the function.
You cannot really convert pointer to member functions to anything and in particular not tovoid* because they have a different sizeof.
If the function varies slightly in signature, you cannot convert from one to the other.For example, even converting from void (MyClass::*)(int) const tovoid (MyClass::*)(int) is not allowed.(You could do it with reinterpret_cast; but that would be an undefined behaviour if you callthem, according to the standard)
Pointer to member functions are not just like normal function pointers.A normal function pointer is just a normal pointer the address where thecode of that function lies.But pointer to member function need to store more information:member functions can be virtual and there is also an offset to apply to thehidden this in case of multiple inheritance.sizeof of a pointer to a member function can evenvary depending of the class.This is why we need to take special care when manipulating them.
Type Traits: QtPrivate::FunctionPointer
Let me introduce you to the QtPrivate::FunctionPointer type trait.
A trait is basically a helper class that gives meta data about a given type.Another example of trait in Qt isQTypeInfo.
What we will need to know in order to implement the new syntax is information about a function pointer.
The template struct FunctionPointer will give us informationabout T via its member.
ArgumentCount: An integer representing the number of arguments of the function.Object: Exists only for pointer to member function. It is a typedef to the class of which the function is a member.Arguments: Represents the list of argument. It is a typedef to a meta-programming list.call(T &function, QObject *receiver, void **args): A static function that will call the function, applying the given parameters.
Qt still supports C++98 compiler which means we unfortunately cannot require support for variadic templates.Therefore we had to specialize our trait function for each number of arguments.We have four kinds of specializationd: normal function pointer, pointer to member function,pointer to const member function and functors.For each kind, we need to specialize for each number of arguments. We support up to six arguments.We also made a specialization using variadic templateso we support arbitrary number of arguments if the compiler supports variadic templates.
The implementation of FunctionPointer lies inqobjectdefs_impl.h.
QObject::connect
The implementation relies on a lot of template code. I am not going to explain all of it.
Here is the code of the first new overload fromqobject.h:
You notice in the function signature that sender and receiverare not just QObject* as the documentation points out. They are pointers totypename FunctionPointer::Object instead.This uses SFINAEto make this overload only enabled for pointers to member functionsbecause the Object only exists in FunctionPointer ifthe type is a pointer to member function.
We then start with a bunch ofQ_STATIC_ASSERT.They should generate sensible compilation error messages when the user made a mistake.If the user did something wrong, it is important that he/she sees an error hereand not in the soup of template code in the _impl.h files.We want to hide the underlying implementation from the user who should not needto care about it.
That means that if you ever you see a confusing error in the implementation details,it should be considered as a bug that should be reported.
We then allocate a QSlotObject that is going to be passed to connectImpl().The QSlotObject is a wrapper around the slot that will help calling it. It alsoknows the type of the signal arguments so it can do the proper type conversion.
We use List_Left to only pass the same number as argument as the slot, which allows connectinga signal with many arguments to a slot with less arguments.
QObject::connectImpl is the private internal functionthat will perform the connection.It is similar to the original syntax, the difference is that instead of storing amethod index in the QObjectPrivate::Connection structure,we store a pointer to the QSlotObjectBase.
The reason why we pass &slot as a void** is only tobe able to compare it if the type is Qt::UniqueConnection.
We also pass the &signal as a void**.It is a pointer to the member function pointer. (Yes, a pointer to the pointer)
Signal Index
We need to make a relationship between the signal pointer and the signal index.
We use MOC for that. Yes, that means this new syntaxis still using the MOC and that there are no plans to get rid of it :-).
MOC will generate code in qt_static_metacallthat compares the parameter and returns the right index.connectImpl will call the qt_static_metacall function with thepointer to the function pointer.
Once we have the signal index, we can proceed like in the other syntax.
The QSlotObjectBase
QSlotObjectBase is the object passed to connectImplthat represents the slot.
Before showing the real code, this is what QObject::QSlotObjectBasewas in Qt5 alpha:
It is basically an interface that is meant to be re-implemented bytemplate classes implementing the call and comparison of thefunction pointers.
It is re-implemented by one of the QSlotObject, QStaticSlotObject orQFunctorSlotObject template class.
Fake Virtual Table
The problem with that is that each instantiation of those object would need to create a virtual table which contains not only pointer to virtual functionsbut also lot of information we do not need such asRTTI.That would result in lot of superfluous data and relocation in the binaries.
In order to avoid that, QSlotObjectBase was changed not to be a C++ polymorphic class.Virtual functions are emulated by hand.
The m_impl is a (normal) function pointer which performsthe three operations that were previously virtual functions. The 're-implementations'set it to their own implementation in the constructor.
Please do not go in your code and replace all your virtual functions by such ahack because you read here it was good.This is only done in this case because almost every call to connectwould generate a new different type (since the QSlotObject has template parameterswich depend on signature of the signal and the slot).
Protected, Public, or Private Signals.
Signals were protected in Qt4 and before. It was a design choice as signals should be emittedby the object when its change its state. They should not be emitted fromoutside the object and calling a signal on another object is almost always a bad idea.
However, with the new syntax, you need to be able take the addressof the signal from the point you make the connection.The compiler would only let you do that if you have access to that signal.Writing &Counter::valueChanged would generate a compiler errorif the signal was not public.
In Qt 5 we had to change signals from protected to public.This is unfortunate since this mean anyone can emit the signals.We found no way around it. We tried a trick with the emit keyword. We tried returning a special value.But nothing worked.I believe that the advantages of the new syntax overcome the problem that signals are now public.
Sometimes it is even desirable to have the signal private. This is the case for example inQAbstractItemModel, where otherwise, developers tend to emit signalfrom the derived class which is not what the API wants.There used to be a pre-processor trick that made signals privatebut it broke the new connection syntax.
A new hack has been introduced.QPrivateSignal is a dummy (empty) struct declared private in the Q_OBJECTmacro. It can be used as the last parameter of the signal. Because it is private, only the objecthas the right to construct it for calling the signal.MOC will ignore the QPrivateSignal last argument while generating signature information.See qabstractitemmodel.h for an example.
More Template Code
The rest of the code is inqobjectdefs_impl.h andqobject_impl.h.It is mostly standard dull template code.
I will not go into much more details in this article,but I will just go over few items that are worth mentioning.
Meta-Programming List
As pointed out earlier, FunctionPointer::Arguments is a listof the arguments. The code needs to operate on that list:iterate over each element, take only a part of it or select a given item.
That is why there isQtPrivate::List that can represent a list of types. Some helpers to operate on it areQtPrivate::List_Select andQtPrivate::List_Left, which give the N-th element in the list and a sub-list containingthe N first elements.
The implementation of List is different for compilers that support variadic templates and compilers that do not.
With variadic templates, it is atemplate struct List;. The list of arguments is just encapsulatedin the template parameters.
For example: the type of a list containing the arguments (int, QString, QObject*) would simply be:
Without variadic template, it is a LISP-style list: template struct List;where Tail can be either another List or void for the end of the list.
The same example as before would be:
ApplyReturnValue Trick
Example Slot Signal Device
In the function FunctionPointer::call, the args[0] is meant to receive the return value of the slot.If the signal returns a value, it is a pointer to an object of the return type ofthe signal, else, it is 0.If the slot returns a value, we need to copy it in arg[0]. If it returns void, we do nothing.
The problem is that it is not syntaxically correct to use thereturn value of a function that returns void.Should I have duplicated the already huge amount of code duplication: once for the voidreturn type and the other for the non-void?No, thanks to the comma operator.
In C++ you can do something like that:
You could have replaced the comma by a semicolon and everything would have been fine.
Where it becomes interesting is when you call it with something that is not void:
There, the comma will actually call an operator that you even can overload.It is what we do inqobjectdefs_impl.h
ApplyReturnValue is just a wrapper around a void*. Then it can be usedin each helper. This is for example the case of a functor without arguments:
This code is inlined, so it will not cost anything at run-time.
Conclusion
This is it for this blog post. There is still a lot to talk about(I have not even mentioned QueuedConnection or thread safety yet), but I hope you found thisinterresting and that you learned here something that might help you as a programmer.
Update:The part 3 is available.
Home > Articles > Programming > Python
␡- Signals and Slots
This chapter is from the book
This chapter is from the book
Signals and Slots
Every GUI library provides the details of events that take place, such as mouse clicks and key presses. For example, if we have a button with the text Click Me, and the user clicks it, all kinds of information becomes available. The GUI library can tell us the coordinates of the mouse click relative to the button, relative to the button's parent widget, and relative to the screen; it can tell us the state of the Shift, Ctrl, Alt, and NumLock keys at the time of the click; and the precise time of the click and of the release; and so on. Similar information can be provided if the user 'clicked' the button without using the mouse. The user may have pressed the Tab key enough times to move the focus to the button and then pressed Spacebar, or maybe they pressed Alt+C. Although the outcome is the same in all these cases, each different means of clicking the button produces different events and different information.
The Qt library was the first to recognize that in almost every case, programmers don't need or even want all the low-level details: They don't care how the button was pressed, they just want to know that it was pressed so that they can respond appropriately. For this reason Qt, and therefore PyQt, provides two communication mechanisms: a low-level event-handling mechanism which is similar to those provided by all the other GUI libraries, and a high-level mechanism which Trolltech (makers of Qt) have called 'signals and slots'. We will look at the low-level mechanism in Chapter 10, and again in Chapter 11, but in this section we will focus on the high-level mechanism.
Every QObject—including all of PyQt's widgets since they derive from QWidget, a QObject subclass—supports the signals and slots mechanism. In particular, they are capable of announcing state changes, such as when a checkbox becomes checked or unchecked, and other important occurrences, for example when a button is clicked (by whatever means). All of PyQt's widgets have a set of predefined signals.
Whenever a signal is emitted, by default PyQt simply throws it away! To take notice of a signal we must connect it to a slot. In C++/Qt, slots are methods that must be declared with a special syntax; but in PyQt, they can be any callable we like (e.g., any function or method), and no special syntax is required when defining them.
Most widgets also have predefined slots, so in some cases we can connect a predefined signal to a predefined slot and not have to do anything else to get the behavior we want. PyQt is more versatile than C++/Qt in this regard, because we can connect not just to slots, but also to any callable, and from PyQt 4.2, it is possible to dynamically add 'predefined' signals and slots to QObjects. Let's see how signals and slots works in practice with the Signals and Slots program shown in Figure 4.6.
Both the QDial and QSpinBox widgets have valueChanged() signals that, when emitted, carry the new value. And they both have setValue() slots that take an integer value. We can therefore connect these two widgets to each other so that whichever one the user changes, will cause the other to be changed correspondingly:
Since the two widgets are connected in this way, if the user moves the dial—say to value 20—the dial will emit a valueChanged(20) signal which will, in turn, cause a call to the spinbox's setValue() slot with 20 as the argument. But then, since its value has now been changed, the spinbox will emit a valueChanged(20) signal which will in turn cause a call to the dial's setValue() slot with 20 as the argument. So it looks like we will get an infinite loop. But what happens is that the valueChanged() signal is not emitted if the value is not actually changed. This is because the standard approach to writing value-changing slots is to begin by comparing the new value with the existing one. If the values are the same, we do nothing and return; otherwise, we apply the change and emit a signal to announce the change of state. The connections are depicted in Figure 4.7.
Figure 4.7 The signals and slots connections
Now let's look at the general syntax for connections. We assume that the PyQt modules have been imported using the from ... import * syntax, and that s and w are QObjects, normally widgets, with s usually being self.
The signalSignature is the name of the signal and a (possibly empty) comma-separated list of parameter type names in parentheses. If the signal is a Qt signal, the type names must be the C++ type names, such as int and QString. C++ type names can be rather complex, with each type name possibly including one or more of const, *, and &. When we write them as signal (or slot) signatures we can drop any consts and &s, but must keep any *s. For example, almost every Qt signal that passes a QString uses a parameter type of const QString&, but in PyQt, just using QString alone is sufficient. On the other hand, the QListWidget has a signal with the signature itemActivated(QListWidgetItem*), and we must use this exactly as written.
PyQt signals are defined when they are actually emitted and can have any number of any type of parameters, as we will see shortly.
The slotSignature has the same form as a signalSignature except that the name is of a Qt slot. A slot may not have more arguments than the signal that is connected to it, but may have less; the additional parameters are then discarded. Corresponding signal and slot arguments must have the same types, so for example, we could not connect a QDial's valueChanged(int) signal to a QLineEdit's setText(QString) slot.
In our dial and spinbox example we used the instance.methodName syntax as we did with the example applications shown earlier in the chapter. But when the slot is actually a Qt slot rather than a Python method, it is more efficient to use the SLOT() syntax:
We have already seen that it is possible to connect multiple signals to the same slot. It is also possible to connect a single signal to multiple slots. Although rare, we can also connect a signal to another signal: In such cases, when the first signal is emitted, it will cause the signal it is connected to, to be emitted.
Connections are made using QObject.connect(); they can be broken using QObject.disconnect(). In practice, we rarely need to break connections ourselves since, for example, PyQt will automatically disconnect any connections involving an object that has been deleted.
So far we have seen how to connect to signals, and how to write slots—which are ordinary functions or methods. And we know that signals are emitted to signify state changes or other important occurrences. But what if we want to create a component that emits its own signals? This is easily achieved using QObject.emit(). For example, here is a complete QSpinBox subclass that emits its own custom atzero signal, and that also passes a number:
We connect to the spinbox's own valueChanged() signal and have it call our checkzero() slot. If the value happens to be 0, the checkzero() slot emits the atzero signal, along with a count of how many times it has been zero; passing additional data like this is optional. The lack of parentheses for the signal is important: It tells PyQt that this is a 'short-circuit' signal.
A signal with no arguments (and therefore no parentheses) is a short-circuit Python signal. When such a signal is emitted, any data can be passed as additional arguments to the emit() method, and they are passed as Python objects. This avoids the overhead of converting the arguments to and from C++ data types, and also means that arbitrary Python objects can be passed, even ones which cannot be converted to and from C++ data types. A signal with at least one argument is either a Qt signal or a non-short-circuit Python signal. In these cases, PyQt will check to see whether the signal is a Qt signal, and if it is not will assume that it is a Python signal. In either case, the arguments are converted to C++ data types.
Here is how we connect to the signal in the form's __init__() method:
Again, we must not use parentheses because it is a short-circuit signal. And for completeness, here is the slot it connects to in the form:
If we use the SIGNAL() function with an identifier but no parentheses, we are specifying a short-circuit signal as described earlier. We can use this syntax both to emit short-circuit signals, and to connect to them. Both uses are shown in the example.
If we use the SIGNAL() function with a signalSignature (a possibly empty parenthesized list of comma-separated PyQt types), we are specifying either a Python or a Qt signal. (A Python signal is one that is emitted in Python code; a Qt signal is one emitted from an underlying C++ object.) We can use this syntax both to emit Python and Qt signals, and to connect to them. These signals can be connected to any callable, that is, to any function or method, including Qt slots; they can also be connected using the SLOT() syntax, with a slotSignature. PyQt checks to see whether the signal is a Qt signal, and if it is not it assumes it is a Python signal. If we use parentheses, even for Python signals, the arguments must be convertible to C++ data types.
We will now look at another example, a tiny custom non-GUI class that has a signal and a slot and which shows that the mechanism is not limited to GUI classes—any QObject subclass can use signals and slots.
Both the rate() and the setRate() methods can be connected to, since any Python callable can be used as a slot. If the rate is changed, we update the private __rate value and emit a custom rateChanged signal, giving the new rate as a parameter. We have also used the faster short-circuit syntax. If we wanted to use the standard syntax, the only difference would be that the signal would be written as SIGNAL('rateChanged(float)'). If we connect the rateChanged signal to the setRate() slot, because of the if statement, no infinite loop will occur. Let us look at the class in use. First we will declare a function to be called when the rate changes:
And now we will try it out:
This will cause just one line to be output to the console: 'TaxRate changed to 8.50%'.
In earlier examples where we connected multiple signals to the same slot, we did not care who emitted the signal. But sometimes we want to connect two or more signals to the same slot, and have the slot behave differently depending on who called it. In this section's last example we will address this issue.
The Connections program shown in Figure 4.8, has five buttons and a label. When one of the buttons is clicked the signals and slots mechanism is used to update the label's text. Here is how the first button is created in the form's __init__() method:
All the other buttons are created in the same way, differing only in their variable name and the text that is passed to them.
We will start with the simplest connection, which is used by button1. Here is the __init__() method's connect() call:
We have used a dedicated method for this button:
Connecting a button's clicked() signal to a single method that responds appropriately is probably the most common connection scenario.
Example Slot Signal Circuit
But what if most of the processing was the same, with just some parameterization depending on which particular button was pressed? In such cases, it is usually best to connect each button to the same slot. There are two approaches to doing this. One is to use partial function application to wrap a slot with a parameter so that when the slot is invoked it is parameterized with the button that called it. The other is to ask PyQt to tell us which button called the slot. We will show both approaches, starting with partial function application.
Back on page 65 we created a wrapper function which used Python 2.5's functools.partial() function or our own simple partial() function:
Using partial() we can now wrap a slot and a button name together. So we might be tempted to do this:
Unfortunately, this won't work for PyQt versions prior to 4.3. The wrapper function is created in the connect() call, but as soon as the connect() call completes, the wrapper goes out of scope and is garbage-collected. From PyQt 4.3, wrappers made with functools.partial() are treated specially when they are used for connections like this. This means that the function connected to will not be garbage-collected, so the code shown earlier will work correctly.
For PyQt 4.0, 4.1, and 4.2, we can still use partial(): We just need to keep a reference to the wrapper—we will not use the reference except for the connect() call, but the fact that it is an attribute of the form instance will ensure that the wrapper function will not go out of scope while the form exists, and will therefore work. So the connection is actually made like this:
When button2 is clicked, the anyButton() method will be called with a string parameter containing the text 'Two'. Here is what this method looks like:
We could have used this slot for all the buttons using the partial() function that we have just shown. And in fact, we could avoid using partial() at all and get the same results:
Example Slot Signal Bar
Here we've created a lambda function that is parameterized by the button's name. It works the same as the partial() technique, and calls the same anyButton() method, only with lambda being used to create the wrapper.
Both button2callback() and button3callback() call anyButton(); the only difference between them is that the first passes 'Two' as its parameter and the second passes 'Three'.
If we are using PyQt 4.1.1 or later, and we use lambda callbacks, we don't have to keep a reference to them ourselves. This is because PyQt treats lambda specially when used to create wrappers in a connection. (This is the same special treatment that is expected to be extended to functools.partial() in PyQt 4.3.) For this reason we can use lambda directly in connect() calls. For example:
The wrapping technique works perfectly well, but there is an alternative approach that is slightly more involved, but which may be useful in some cases, particularly when we don't want to wrap our calls. This other technique is used to respond to button4 and to button5. Here are their connections:
Notice that we do not wrap the clicked() method that they are both connected to, so at first sight it looks like there is no way to tell which button called the clicked() method.* However, the implementation makes clear that we can distinguish if we want to:
Inside a slot we can always call sender() to discover which QObject the invoking signal came from. (This could be None if the slot was called using a normal method call.) Although we know that we have connected only buttons to this slot, we still take care to check. We have used isinstance(), but we could have used hasattr(button, 'text') instead. If we had connected all the buttons to this slot, it would have worked correctly for them all.
Some programmers don't like using sender() because they feel that it isn't good object-oriented style, so they tend to use partial function application when needs like this arise.
There is actually one other technique that can be used to get the effect of wrapping a function and a parameter. It makes use of the QSignalMapper class, and an example of its use is shown in Chapter 9.
It is possible in some situations for a slot to be called as the result of a signal, and the processing performed in the slot, directly or indirectly, causes the signal that originally called the slot to be called again, leading to an infinite cycle. Such cycles are rare in practice. Two factors help reduce the possibility of cycles. First, some signals are emitted only if a real change takes place. For example, if the value of a QSpinBox is changed by the user, or programmatically by a setValue() call, it emits its valueChanged() signal only if the new value is different from the current value. Second, some signals are emitted only as the result of user actions. For example, QLineEdit emits its textEdited() signal only when the text is changed by the user, and not when it is changed in code by a setText() call.
If a signal–slot cycle does seem to have occurred, naturally, the first thing to check is that the code's logic is correct: Are we actually doing the processing we thought we were? If the logic is right, and we still have a cycle, we might be able to break the cycle by changing the signals that we connect to—for example, replacing signals that are emitted as a result of programmatic changes, with those that are emitted only as a result of user interaction. If the problem persists, we could stop signals being emitted at certain places in our code using QObject.blockSignals(), which is inherited by all QWidget classes and is passed a Boolean—True to stop the object emitting signals and False to resume signalling.
This completes our formal coverage of the signals and slots mechanism. We will see many more examples of signals and slots in practice in almost all the examples shown in the rest of the book. Most other GUI libraries have copied the mechanism in some form or other. This is because the signals and slots mechanism is very useful and powerful, and leaves programmers free to focus on the logic of their applications rather than having to concern themselves with the details of how the user invoked a particular operation.
Related Resources
- Online Video $239.99
The implementation of FunctionPointer lies inqobjectdefs_impl.h.
QObject::connect
The implementation relies on a lot of template code. I am not going to explain all of it.
Here is the code of the first new overload fromqobject.h:
You notice in the function signature that sender and receiverare not just QObject* as the documentation points out. They are pointers totypename FunctionPointer::Object instead.This uses SFINAEto make this overload only enabled for pointers to member functionsbecause the Object only exists in FunctionPointer ifthe type is a pointer to member function.
We then start with a bunch ofQ_STATIC_ASSERT.They should generate sensible compilation error messages when the user made a mistake.If the user did something wrong, it is important that he/she sees an error hereand not in the soup of template code in the _impl.h files.We want to hide the underlying implementation from the user who should not needto care about it.
That means that if you ever you see a confusing error in the implementation details,it should be considered as a bug that should be reported.
We then allocate a QSlotObject that is going to be passed to connectImpl().The QSlotObject is a wrapper around the slot that will help calling it. It alsoknows the type of the signal arguments so it can do the proper type conversion.
We use List_Left to only pass the same number as argument as the slot, which allows connectinga signal with many arguments to a slot with less arguments.
QObject::connectImpl is the private internal functionthat will perform the connection.It is similar to the original syntax, the difference is that instead of storing amethod index in the QObjectPrivate::Connection structure,we store a pointer to the QSlotObjectBase.
The reason why we pass &slot as a void** is only tobe able to compare it if the type is Qt::UniqueConnection.
We also pass the &signal as a void**.It is a pointer to the member function pointer. (Yes, a pointer to the pointer)
Signal Index
We need to make a relationship between the signal pointer and the signal index.
We use MOC for that. Yes, that means this new syntaxis still using the MOC and that there are no plans to get rid of it :-).
MOC will generate code in qt_static_metacallthat compares the parameter and returns the right index.connectImpl will call the qt_static_metacall function with thepointer to the function pointer.
Once we have the signal index, we can proceed like in the other syntax.
The QSlotObjectBase
QSlotObjectBase is the object passed to connectImplthat represents the slot.
Before showing the real code, this is what QObject::QSlotObjectBasewas in Qt5 alpha:
It is basically an interface that is meant to be re-implemented bytemplate classes implementing the call and comparison of thefunction pointers.
It is re-implemented by one of the QSlotObject, QStaticSlotObject orQFunctorSlotObject template class.
Fake Virtual Table
The problem with that is that each instantiation of those object would need to create a virtual table which contains not only pointer to virtual functionsbut also lot of information we do not need such asRTTI.That would result in lot of superfluous data and relocation in the binaries.
In order to avoid that, QSlotObjectBase was changed not to be a C++ polymorphic class.Virtual functions are emulated by hand.
The m_impl is a (normal) function pointer which performsthe three operations that were previously virtual functions. The 're-implementations'set it to their own implementation in the constructor.
Please do not go in your code and replace all your virtual functions by such ahack because you read here it was good.This is only done in this case because almost every call to connectwould generate a new different type (since the QSlotObject has template parameterswich depend on signature of the signal and the slot).
Protected, Public, or Private Signals.
Signals were protected in Qt4 and before. It was a design choice as signals should be emittedby the object when its change its state. They should not be emitted fromoutside the object and calling a signal on another object is almost always a bad idea.
However, with the new syntax, you need to be able take the addressof the signal from the point you make the connection.The compiler would only let you do that if you have access to that signal.Writing &Counter::valueChanged would generate a compiler errorif the signal was not public.
In Qt 5 we had to change signals from protected to public.This is unfortunate since this mean anyone can emit the signals.We found no way around it. We tried a trick with the emit keyword. We tried returning a special value.But nothing worked.I believe that the advantages of the new syntax overcome the problem that signals are now public.
Sometimes it is even desirable to have the signal private. This is the case for example inQAbstractItemModel, where otherwise, developers tend to emit signalfrom the derived class which is not what the API wants.There used to be a pre-processor trick that made signals privatebut it broke the new connection syntax.
A new hack has been introduced.QPrivateSignal is a dummy (empty) struct declared private in the Q_OBJECTmacro. It can be used as the last parameter of the signal. Because it is private, only the objecthas the right to construct it for calling the signal.MOC will ignore the QPrivateSignal last argument while generating signature information.See qabstractitemmodel.h for an example.
More Template Code
The rest of the code is inqobjectdefs_impl.h andqobject_impl.h.It is mostly standard dull template code.
I will not go into much more details in this article,but I will just go over few items that are worth mentioning.
Meta-Programming List
As pointed out earlier, FunctionPointer::Arguments is a listof the arguments. The code needs to operate on that list:iterate over each element, take only a part of it or select a given item.
That is why there isQtPrivate::List that can represent a list of types. Some helpers to operate on it areQtPrivate::List_Select andQtPrivate::List_Left, which give the N-th element in the list and a sub-list containingthe N first elements.
The implementation of List is different for compilers that support variadic templates and compilers that do not.
With variadic templates, it is atemplate struct List;. The list of arguments is just encapsulatedin the template parameters.
For example: the type of a list containing the arguments (int, QString, QObject*) would simply be:
Without variadic template, it is a LISP-style list: template struct List;where Tail can be either another List or void for the end of the list.
The same example as before would be:
ApplyReturnValue Trick
Example Slot Signal Device
In the function FunctionPointer::call, the args[0] is meant to receive the return value of the slot.If the signal returns a value, it is a pointer to an object of the return type ofthe signal, else, it is 0.If the slot returns a value, we need to copy it in arg[0]. If it returns void, we do nothing.
The problem is that it is not syntaxically correct to use thereturn value of a function that returns void.Should I have duplicated the already huge amount of code duplication: once for the voidreturn type and the other for the non-void?No, thanks to the comma operator.
In C++ you can do something like that:
You could have replaced the comma by a semicolon and everything would have been fine.
Where it becomes interesting is when you call it with something that is not void:
There, the comma will actually call an operator that you even can overload.It is what we do inqobjectdefs_impl.h
ApplyReturnValue is just a wrapper around a void*. Then it can be usedin each helper. This is for example the case of a functor without arguments:
This code is inlined, so it will not cost anything at run-time.
Conclusion
This is it for this blog post. There is still a lot to talk about(I have not even mentioned QueuedConnection or thread safety yet), but I hope you found thisinterresting and that you learned here something that might help you as a programmer.
Update:The part 3 is available.
Home > Articles > Programming > Python
␡- Signals and Slots
This chapter is from the book
This chapter is from the book
Signals and Slots
Every GUI library provides the details of events that take place, such as mouse clicks and key presses. For example, if we have a button with the text Click Me, and the user clicks it, all kinds of information becomes available. The GUI library can tell us the coordinates of the mouse click relative to the button, relative to the button's parent widget, and relative to the screen; it can tell us the state of the Shift, Ctrl, Alt, and NumLock keys at the time of the click; and the precise time of the click and of the release; and so on. Similar information can be provided if the user 'clicked' the button without using the mouse. The user may have pressed the Tab key enough times to move the focus to the button and then pressed Spacebar, or maybe they pressed Alt+C. Although the outcome is the same in all these cases, each different means of clicking the button produces different events and different information.
The Qt library was the first to recognize that in almost every case, programmers don't need or even want all the low-level details: They don't care how the button was pressed, they just want to know that it was pressed so that they can respond appropriately. For this reason Qt, and therefore PyQt, provides two communication mechanisms: a low-level event-handling mechanism which is similar to those provided by all the other GUI libraries, and a high-level mechanism which Trolltech (makers of Qt) have called 'signals and slots'. We will look at the low-level mechanism in Chapter 10, and again in Chapter 11, but in this section we will focus on the high-level mechanism.
Every QObject—including all of PyQt's widgets since they derive from QWidget, a QObject subclass—supports the signals and slots mechanism. In particular, they are capable of announcing state changes, such as when a checkbox becomes checked or unchecked, and other important occurrences, for example when a button is clicked (by whatever means). All of PyQt's widgets have a set of predefined signals.
Whenever a signal is emitted, by default PyQt simply throws it away! To take notice of a signal we must connect it to a slot. In C++/Qt, slots are methods that must be declared with a special syntax; but in PyQt, they can be any callable we like (e.g., any function or method), and no special syntax is required when defining them.
Most widgets also have predefined slots, so in some cases we can connect a predefined signal to a predefined slot and not have to do anything else to get the behavior we want. PyQt is more versatile than C++/Qt in this regard, because we can connect not just to slots, but also to any callable, and from PyQt 4.2, it is possible to dynamically add 'predefined' signals and slots to QObjects. Let's see how signals and slots works in practice with the Signals and Slots program shown in Figure 4.6.
Both the QDial and QSpinBox widgets have valueChanged() signals that, when emitted, carry the new value. And they both have setValue() slots that take an integer value. We can therefore connect these two widgets to each other so that whichever one the user changes, will cause the other to be changed correspondingly:
Since the two widgets are connected in this way, if the user moves the dial—say to value 20—the dial will emit a valueChanged(20) signal which will, in turn, cause a call to the spinbox's setValue() slot with 20 as the argument. But then, since its value has now been changed, the spinbox will emit a valueChanged(20) signal which will in turn cause a call to the dial's setValue() slot with 20 as the argument. So it looks like we will get an infinite loop. But what happens is that the valueChanged() signal is not emitted if the value is not actually changed. This is because the standard approach to writing value-changing slots is to begin by comparing the new value with the existing one. If the values are the same, we do nothing and return; otherwise, we apply the change and emit a signal to announce the change of state. The connections are depicted in Figure 4.7.
Figure 4.7 The signals and slots connections
Now let's look at the general syntax for connections. We assume that the PyQt modules have been imported using the from ... import * syntax, and that s and w are QObjects, normally widgets, with s usually being self.
The signalSignature is the name of the signal and a (possibly empty) comma-separated list of parameter type names in parentheses. If the signal is a Qt signal, the type names must be the C++ type names, such as int and QString. C++ type names can be rather complex, with each type name possibly including one or more of const, *, and &. When we write them as signal (or slot) signatures we can drop any consts and &s, but must keep any *s. For example, almost every Qt signal that passes a QString uses a parameter type of const QString&, but in PyQt, just using QString alone is sufficient. On the other hand, the QListWidget has a signal with the signature itemActivated(QListWidgetItem*), and we must use this exactly as written.
PyQt signals are defined when they are actually emitted and can have any number of any type of parameters, as we will see shortly.
The slotSignature has the same form as a signalSignature except that the name is of a Qt slot. A slot may not have more arguments than the signal that is connected to it, but may have less; the additional parameters are then discarded. Corresponding signal and slot arguments must have the same types, so for example, we could not connect a QDial's valueChanged(int) signal to a QLineEdit's setText(QString) slot.
In our dial and spinbox example we used the instance.methodName syntax as we did with the example applications shown earlier in the chapter. But when the slot is actually a Qt slot rather than a Python method, it is more efficient to use the SLOT() syntax:
We have already seen that it is possible to connect multiple signals to the same slot. It is also possible to connect a single signal to multiple slots. Although rare, we can also connect a signal to another signal: In such cases, when the first signal is emitted, it will cause the signal it is connected to, to be emitted.
Connections are made using QObject.connect(); they can be broken using QObject.disconnect(). In practice, we rarely need to break connections ourselves since, for example, PyQt will automatically disconnect any connections involving an object that has been deleted.
So far we have seen how to connect to signals, and how to write slots—which are ordinary functions or methods. And we know that signals are emitted to signify state changes or other important occurrences. But what if we want to create a component that emits its own signals? This is easily achieved using QObject.emit(). For example, here is a complete QSpinBox subclass that emits its own custom atzero signal, and that also passes a number:
We connect to the spinbox's own valueChanged() signal and have it call our checkzero() slot. If the value happens to be 0, the checkzero() slot emits the atzero signal, along with a count of how many times it has been zero; passing additional data like this is optional. The lack of parentheses for the signal is important: It tells PyQt that this is a 'short-circuit' signal.
A signal with no arguments (and therefore no parentheses) is a short-circuit Python signal. When such a signal is emitted, any data can be passed as additional arguments to the emit() method, and they are passed as Python objects. This avoids the overhead of converting the arguments to and from C++ data types, and also means that arbitrary Python objects can be passed, even ones which cannot be converted to and from C++ data types. A signal with at least one argument is either a Qt signal or a non-short-circuit Python signal. In these cases, PyQt will check to see whether the signal is a Qt signal, and if it is not will assume that it is a Python signal. In either case, the arguments are converted to C++ data types.
Here is how we connect to the signal in the form's __init__() method:
Again, we must not use parentheses because it is a short-circuit signal. And for completeness, here is the slot it connects to in the form:
If we use the SIGNAL() function with an identifier but no parentheses, we are specifying a short-circuit signal as described earlier. We can use this syntax both to emit short-circuit signals, and to connect to them. Both uses are shown in the example.
If we use the SIGNAL() function with a signalSignature (a possibly empty parenthesized list of comma-separated PyQt types), we are specifying either a Python or a Qt signal. (A Python signal is one that is emitted in Python code; a Qt signal is one emitted from an underlying C++ object.) We can use this syntax both to emit Python and Qt signals, and to connect to them. These signals can be connected to any callable, that is, to any function or method, including Qt slots; they can also be connected using the SLOT() syntax, with a slotSignature. PyQt checks to see whether the signal is a Qt signal, and if it is not it assumes it is a Python signal. If we use parentheses, even for Python signals, the arguments must be convertible to C++ data types.
We will now look at another example, a tiny custom non-GUI class that has a signal and a slot and which shows that the mechanism is not limited to GUI classes—any QObject subclass can use signals and slots.
Both the rate() and the setRate() methods can be connected to, since any Python callable can be used as a slot. If the rate is changed, we update the private __rate value and emit a custom rateChanged signal, giving the new rate as a parameter. We have also used the faster short-circuit syntax. If we wanted to use the standard syntax, the only difference would be that the signal would be written as SIGNAL('rateChanged(float)'). If we connect the rateChanged signal to the setRate() slot, because of the if statement, no infinite loop will occur. Let us look at the class in use. First we will declare a function to be called when the rate changes:
And now we will try it out:
This will cause just one line to be output to the console: 'TaxRate changed to 8.50%'.
In earlier examples where we connected multiple signals to the same slot, we did not care who emitted the signal. But sometimes we want to connect two or more signals to the same slot, and have the slot behave differently depending on who called it. In this section's last example we will address this issue.
The Connections program shown in Figure 4.8, has five buttons and a label. When one of the buttons is clicked the signals and slots mechanism is used to update the label's text. Here is how the first button is created in the form's __init__() method:
All the other buttons are created in the same way, differing only in their variable name and the text that is passed to them.
We will start with the simplest connection, which is used by button1. Here is the __init__() method's connect() call:
We have used a dedicated method for this button:
Connecting a button's clicked() signal to a single method that responds appropriately is probably the most common connection scenario.
Example Slot Signal Circuit
But what if most of the processing was the same, with just some parameterization depending on which particular button was pressed? In such cases, it is usually best to connect each button to the same slot. There are two approaches to doing this. One is to use partial function application to wrap a slot with a parameter so that when the slot is invoked it is parameterized with the button that called it. The other is to ask PyQt to tell us which button called the slot. We will show both approaches, starting with partial function application.
Back on page 65 we created a wrapper function which used Python 2.5's functools.partial() function or our own simple partial() function:
Using partial() we can now wrap a slot and a button name together. So we might be tempted to do this:
Unfortunately, this won't work for PyQt versions prior to 4.3. The wrapper function is created in the connect() call, but as soon as the connect() call completes, the wrapper goes out of scope and is garbage-collected. From PyQt 4.3, wrappers made with functools.partial() are treated specially when they are used for connections like this. This means that the function connected to will not be garbage-collected, so the code shown earlier will work correctly.
For PyQt 4.0, 4.1, and 4.2, we can still use partial(): We just need to keep a reference to the wrapper—we will not use the reference except for the connect() call, but the fact that it is an attribute of the form instance will ensure that the wrapper function will not go out of scope while the form exists, and will therefore work. So the connection is actually made like this:
When button2 is clicked, the anyButton() method will be called with a string parameter containing the text 'Two'. Here is what this method looks like:
We could have used this slot for all the buttons using the partial() function that we have just shown. And in fact, we could avoid using partial() at all and get the same results:
Example Slot Signal Bar
Here we've created a lambda function that is parameterized by the button's name. It works the same as the partial() technique, and calls the same anyButton() method, only with lambda being used to create the wrapper.
Both button2callback() and button3callback() call anyButton(); the only difference between them is that the first passes 'Two' as its parameter and the second passes 'Three'.
If we are using PyQt 4.1.1 or later, and we use lambda callbacks, we don't have to keep a reference to them ourselves. This is because PyQt treats lambda specially when used to create wrappers in a connection. (This is the same special treatment that is expected to be extended to functools.partial() in PyQt 4.3.) For this reason we can use lambda directly in connect() calls. For example:
The wrapping technique works perfectly well, but there is an alternative approach that is slightly more involved, but which may be useful in some cases, particularly when we don't want to wrap our calls. This other technique is used to respond to button4 and to button5. Here are their connections:
Notice that we do not wrap the clicked() method that they are both connected to, so at first sight it looks like there is no way to tell which button called the clicked() method.* However, the implementation makes clear that we can distinguish if we want to:
Inside a slot we can always call sender() to discover which QObject the invoking signal came from. (This could be None if the slot was called using a normal method call.) Although we know that we have connected only buttons to this slot, we still take care to check. We have used isinstance(), but we could have used hasattr(button, 'text') instead. If we had connected all the buttons to this slot, it would have worked correctly for them all.
Some programmers don't like using sender() because they feel that it isn't good object-oriented style, so they tend to use partial function application when needs like this arise.
There is actually one other technique that can be used to get the effect of wrapping a function and a parameter. It makes use of the QSignalMapper class, and an example of its use is shown in Chapter 9.
It is possible in some situations for a slot to be called as the result of a signal, and the processing performed in the slot, directly or indirectly, causes the signal that originally called the slot to be called again, leading to an infinite cycle. Such cycles are rare in practice. Two factors help reduce the possibility of cycles. First, some signals are emitted only if a real change takes place. For example, if the value of a QSpinBox is changed by the user, or programmatically by a setValue() call, it emits its valueChanged() signal only if the new value is different from the current value. Second, some signals are emitted only as the result of user actions. For example, QLineEdit emits its textEdited() signal only when the text is changed by the user, and not when it is changed in code by a setText() call.
If a signal–slot cycle does seem to have occurred, naturally, the first thing to check is that the code's logic is correct: Are we actually doing the processing we thought we were? If the logic is right, and we still have a cycle, we might be able to break the cycle by changing the signals that we connect to—for example, replacing signals that are emitted as a result of programmatic changes, with those that are emitted only as a result of user interaction. If the problem persists, we could stop signals being emitted at certain places in our code using QObject.blockSignals(), which is inherited by all QWidget classes and is passed a Boolean—True to stop the object emitting signals and False to resume signalling.
This completes our formal coverage of the signals and slots mechanism. We will see many more examples of signals and slots in practice in almost all the examples shown in the rest of the book. Most other GUI libraries have copied the mechanism in some form or other. This is because the signals and slots mechanism is very useful and powerful, and leaves programmers free to focus on the logic of their applications rather than having to concern themselves with the details of how the user invoked a particular operation.
Related Resources
- Online Video $239.99
- eBook (Watermarked) $31.99
- Book $39.99
