I recently heard the term "hook" while talking to some people about a program I was writing. I'm unsure exactly what this term implies although I inferred from the conversation that a hook is a type of function. I searched for a definition but was unable to find a good answer. Would someone be able to give me an idea of what this term generally means and perhaps a small example to illustrate the definition?
This question is related to
hook
Oftentimes hooking refers to Win32 message hooking or the Linux/OSX equivalents, but more generically hooking is simply notifying another object/window/program/etc that you want to be notified when a specified action happens. For instance: Having all windows on the system notify you as they are about to close.
As a general rule, hooking is somewhat hazardous since doing it without understanding how it affects the system can lead to instability or at the very leas unexpected behaviour. It can also be VERY useful in certain circumstances, thought. For instance: FRAPS uses it to determine which windows it should show it's FPS counter on.
Hooks are a category of function that allows base code to call extension code. This can be useful in situations in which a core developer wants to offer extensibility without exposing their code.
One usage of hooks is in video game mod development. A game may not allow mod developers to extend base functionality, but hooks can be added by core mod library developers. With these hooks, independent developers can have their custom code called upon any desired event, such as game loading, inventory updates, entity interactions, etc.
A common method of implementation is to give a function an empty list of callbacks, then expose the ability to extend the list of callbacks. The base code will always call the function at the same and proper time but, with an empty callback list, the function does nothing. This is by design.
A third party, then, has the opportunity to write additional code and add their new callback to the hook's callback list. With nothing more than a reference of available hooks, they have extended functionality at minimal risk to the base system.
Hooks don't allow developers to do anything that can't be done with other structures and interfaces. They are a choice to be made with consideration to the task and users (third-party developers).
For clarification: a hook allows the extension and may be implemented using callbacks. Callbacks are generally nothing more than a function pointer; the computed address of a function. There appears to be confusion in other answers/comments.
hooks can be executed when some condition is encountered. e.g. some variable changes or some action is called or some event happens. hooks can enter in the process and change things or react upon changes.
Oftentimes hooking refers to Win32 message hooking or the Linux/OSX equivalents, but more generically hooking is simply notifying another object/window/program/etc that you want to be notified when a specified action happens. For instance: Having all windows on the system notify you as they are about to close.
As a general rule, hooking is somewhat hazardous since doing it without understanding how it affects the system can lead to instability or at the very leas unexpected behaviour. It can also be VERY useful in certain circumstances, thought. For instance: FRAPS uses it to determine which windows it should show it's FPS counter on.
A hook is functionality provided by software for users of that software to have their own code called under certain circumstances. That code can augment or replace the current code.
In the olden days when computers were truly personal and viruses were less prevalent (I'm talking the '80's), it was as simple as patching the operating system software itself to call your code. I remember writing an extension to the Applesoft BASIC language on the Apple II which simply hooked my code into the BASIC interpreter by injecting a call to my code before any of the line was processed.
Some computers had pre-designed hooks, one example being the I/O stream on the Apple II. It used such a hook to inject the whole disk sub-system (Apple II ROMs were originally built in the days where cassettes were the primary storage medium for PCs). You controlled the disks by printing the ASCII code 4 (CTRL-D) followed by the command you wanted to execute then a CR, and it was intercepted by the disk sub-system, which had hooked itself into the Apple ROM print routines.
So for example, the lines:
PRINT CHR(4);"CATALOG"
PRINT CHR(4);"IN#6"
would list the disk contents then re-initialize the machine. This allowed such tricks as protecting your BASIC programs by setting the first line as:
123 REM XIN#6
then using POKE to insert the CTRL-D character in where the X was. Then, anyone trying to list your source would send the re-initialize sequence through the output routines where the disk sub-system would detect it.
That's often the sort of trickery we had to resort to, to get the behavior we wanted.
Nowadays, with the operating system more secure, it provides facilities for hooks itself, since you're no longer supposed to modify the operating system "in-flight" or on the disk.
They've been around for a long time. Mainframes had them (called exits) and a great deal of mainframe software uses those facilities even now. For example, the free source code control system that comes with z/OS (called SCLM) allows you to entirely replace the security subsystem by simply placing your own code in the exit.
In a generic sense, a "hook" is something that will let you, a programmer, view and/or interact with and/or change something that's already going on in a system/program.
For example, the Drupal CMS provides developers with hooks that let them take additional action after a "content node" is created. If a developer doesn't implement a hook, the node is created per normal. If a developer implements a hook, they can have some additional code run whenever a node is created. This code could do anything, including rolling back and/or altering the original action. It could also do something unrelated to the node creation entirely.
A callback could be thought of as a specific kind of hook. By implementing callback functionality into a system, that system is letting you call some additional code after an action has completed. However, hooking (as a generic term) is not limited to callbacks.
Another example. Sometimes Web Developers will refer to class names and/or IDs on elements as hooks. That's because by placing the ID/class name on an element, they can then use Javascript to modify that element, or "hook in" to the page document. (this is stretching the meaning, but it is commonly used and worth mentioning)
A chain of hooks is a set of functions in which each function calls the next. What is significant about a chain of hooks is that a programmer can add another function to the chain at run time. One way to do this is to look for a known location where the address of the first function in a chain is kept. You then save the value of that function pointer and overwrite the value at the initial address with the address of the function you wish to insert into the hook chain. The function then gets called, does its business and calls the next function in the chain (unless you decide otherwise). Naturally, there are a number of other ways to create a chain of hooks, from writing directly to memory to using the metaprogramming facilities of languages like Ruby or Python.
An example of a chain of hooks is the way that an MS Windows application processes messages. Each function in the processing chain either processes a message or sends it to the next function in the chain.
In the Drupal content management system, 'hook' has a relatively specific meaning. When an internal event occurs (like content creation or user login, for example), modules can respond to the event by implementing a special "hook" function. This is done via naming convention -- [your-plugin-name]_user_login() for the User Login event, for example.
Because of this convention, the underlying events are referred to as "hooks" and appear with names like "hook_user_login" and "hook_user_authenticate()" in Drupal's API documentation.
hooks can be executed when some condition is encountered. e.g. some variable changes or some action is called or some event happens. hooks can enter in the process and change things or react upon changes.
Hooking in programming is a technique employing so-called hooks to make a chain of procedures as an event handler.
Oftentimes hooking refers to Win32 message hooking or the Linux/OSX equivalents, but more generically hooking is simply notifying another object/window/program/etc that you want to be notified when a specified action happens. For instance: Having all windows on the system notify you as they are about to close.
As a general rule, hooking is somewhat hazardous since doing it without understanding how it affects the system can lead to instability or at the very leas unexpected behaviour. It can also be VERY useful in certain circumstances, thought. For instance: FRAPS uses it to determine which windows it should show it's FPS counter on.
In VERY short, you can change the code of an API call such as MessageBox to where it does a different function edited by you (globally will work system wide, locally will work process wide).
Hooking in programming is a technique employing so-called hooks to make a chain of procedures as an event handler.
In the Drupal content management system, 'hook' has a relatively specific meaning. When an internal event occurs (like content creation or user login, for example), modules can respond to the event by implementing a special "hook" function. This is done via naming convention -- [your-plugin-name]_user_login() for the User Login event, for example.
Because of this convention, the underlying events are referred to as "hooks" and appear with names like "hook_user_login" and "hook_user_authenticate()" in Drupal's API documentation.
A chain of hooks is a set of functions in which each function calls the next. What is significant about a chain of hooks is that a programmer can add another function to the chain at run time. One way to do this is to look for a known location where the address of the first function in a chain is kept. You then save the value of that function pointer and overwrite the value at the initial address with the address of the function you wish to insert into the hook chain. The function then gets called, does its business and calls the next function in the chain (unless you decide otherwise). Naturally, there are a number of other ways to create a chain of hooks, from writing directly to memory to using the metaprogramming facilities of languages like Ruby or Python.
An example of a chain of hooks is the way that an MS Windows application processes messages. Each function in the processing chain either processes a message or sends it to the next function in the chain.
In the Drupal content management system, 'hook' has a relatively specific meaning. When an internal event occurs (like content creation or user login, for example), modules can respond to the event by implementing a special "hook" function. This is done via naming convention -- [your-plugin-name]_user_login() for the User Login event, for example.
Because of this convention, the underlying events are referred to as "hooks" and appear with names like "hook_user_login" and "hook_user_authenticate()" in Drupal's API documentation.
hooks can be executed when some condition is encountered. e.g. some variable changes or some action is called or some event happens. hooks can enter in the process and change things or react upon changes.
Hook denotes a place in the code where you dispatch an event of certain type, and if this event was registered before with a proper function to call back, then it would be handled by this registered function, otherwise nothing happens.
Simple said:
A hook is a means of executing custom code (function) either before, after, or instead of existing code. For example, a function may be written to "hook" into the login process in order to execute a Captcha function before continuing on to the normal login process.
In VERY short, you can change the code of an API call such as MessageBox to where it does a different function edited by you (globally will work system wide, locally will work process wide).
hooks can be executed when some condition is encountered. e.g. some variable changes or some action is called or some event happens. hooks can enter in the process and change things or react upon changes.
A hook is functionality provided by software for users of that software to have their own code called under certain circumstances. That code can augment or replace the current code.
In the olden days when computers were truly personal and viruses were less prevalent (I'm talking the '80's), it was as simple as patching the operating system software itself to call your code. I remember writing an extension to the Applesoft BASIC language on the Apple II which simply hooked my code into the BASIC interpreter by injecting a call to my code before any of the line was processed.
Some computers had pre-designed hooks, one example being the I/O stream on the Apple II. It used such a hook to inject the whole disk sub-system (Apple II ROMs were originally built in the days where cassettes were the primary storage medium for PCs). You controlled the disks by printing the ASCII code 4 (CTRL-D) followed by the command you wanted to execute then a CR, and it was intercepted by the disk sub-system, which had hooked itself into the Apple ROM print routines.
So for example, the lines:
PRINT CHR(4);"CATALOG"
PRINT CHR(4);"IN#6"
would list the disk contents then re-initialize the machine. This allowed such tricks as protecting your BASIC programs by setting the first line as:
123 REM XIN#6
then using POKE to insert the CTRL-D character in where the X was. Then, anyone trying to list your source would send the re-initialize sequence through the output routines where the disk sub-system would detect it.
That's often the sort of trickery we had to resort to, to get the behavior we wanted.
Nowadays, with the operating system more secure, it provides facilities for hooks itself, since you're no longer supposed to modify the operating system "in-flight" or on the disk.
They've been around for a long time. Mainframes had them (called exits) and a great deal of mainframe software uses those facilities even now. For example, the free source code control system that comes with z/OS (called SCLM) allows you to entirely replace the security subsystem by simply placing your own code in the exit.
Hooking in programming is a technique employing so-called hooks to make a chain of procedures as an event handler.
In the Drupal content management system, 'hook' has a relatively specific meaning. When an internal event occurs (like content creation or user login, for example), modules can respond to the event by implementing a special "hook" function. This is done via naming convention -- [your-plugin-name]_user_login() for the User Login event, for example.
Because of this convention, the underlying events are referred to as "hooks" and appear with names like "hook_user_login" and "hook_user_authenticate()" in Drupal's API documentation.
Hooking in programming is a technique employing so-called hooks to make a chain of procedures as an event handler.
Hook denotes a place in the code where you dispatch an event of certain type, and if this event was registered before with a proper function to call back, then it would be handled by this registered function, otherwise nothing happens.
A hook is functionality provided by software for users of that software to have their own code called under certain circumstances. That code can augment or replace the current code.
In the olden days when computers were truly personal and viruses were less prevalent (I'm talking the '80's), it was as simple as patching the operating system software itself to call your code. I remember writing an extension to the Applesoft BASIC language on the Apple II which simply hooked my code into the BASIC interpreter by injecting a call to my code before any of the line was processed.
Some computers had pre-designed hooks, one example being the I/O stream on the Apple II. It used such a hook to inject the whole disk sub-system (Apple II ROMs were originally built in the days where cassettes were the primary storage medium for PCs). You controlled the disks by printing the ASCII code 4 (CTRL-D) followed by the command you wanted to execute then a CR, and it was intercepted by the disk sub-system, which had hooked itself into the Apple ROM print routines.
So for example, the lines:
PRINT CHR(4);"CATALOG"
PRINT CHR(4);"IN#6"
would list the disk contents then re-initialize the machine. This allowed such tricks as protecting your BASIC programs by setting the first line as:
123 REM XIN#6
then using POKE to insert the CTRL-D character in where the X was. Then, anyone trying to list your source would send the re-initialize sequence through the output routines where the disk sub-system would detect it.
That's often the sort of trickery we had to resort to, to get the behavior we wanted.
Nowadays, with the operating system more secure, it provides facilities for hooks itself, since you're no longer supposed to modify the operating system "in-flight" or on the disk.
They've been around for a long time. Mainframes had them (called exits) and a great deal of mainframe software uses those facilities even now. For example, the free source code control system that comes with z/OS (called SCLM) allows you to entirely replace the security subsystem by simply placing your own code in the exit.
In a generic sense, a "hook" is something that will let you, a programmer, view and/or interact with and/or change something that's already going on in a system/program.
For example, the Drupal CMS provides developers with hooks that let them take additional action after a "content node" is created. If a developer doesn't implement a hook, the node is created per normal. If a developer implements a hook, they can have some additional code run whenever a node is created. This code could do anything, including rolling back and/or altering the original action. It could also do something unrelated to the node creation entirely.
A callback could be thought of as a specific kind of hook. By implementing callback functionality into a system, that system is letting you call some additional code after an action has completed. However, hooking (as a generic term) is not limited to callbacks.
Another example. Sometimes Web Developers will refer to class names and/or IDs on elements as hooks. That's because by placing the ID/class name on an element, they can then use Javascript to modify that element, or "hook in" to the page document. (this is stretching the meaning, but it is commonly used and worth mentioning)
Hooks are a category of function that allows base code to call extension code. This can be useful in situations in which a core developer wants to offer extensibility without exposing their code.
One usage of hooks is in video game mod development. A game may not allow mod developers to extend base functionality, but hooks can be added by core mod library developers. With these hooks, independent developers can have their custom code called upon any desired event, such as game loading, inventory updates, entity interactions, etc.
A common method of implementation is to give a function an empty list of callbacks, then expose the ability to extend the list of callbacks. The base code will always call the function at the same and proper time but, with an empty callback list, the function does nothing. This is by design.
A third party, then, has the opportunity to write additional code and add their new callback to the hook's callback list. With nothing more than a reference of available hooks, they have extended functionality at minimal risk to the base system.
Hooks don't allow developers to do anything that can't be done with other structures and interfaces. They are a choice to be made with consideration to the task and users (third-party developers).
For clarification: a hook allows the extension and may be implemented using callbacks. Callbacks are generally nothing more than a function pointer; the computed address of a function. There appears to be confusion in other answers/comments.
In a generic sense, a "hook" is something that will let you, a programmer, view and/or interact with and/or change something that's already going on in a system/program.
For example, the Drupal CMS provides developers with hooks that let them take additional action after a "content node" is created. If a developer doesn't implement a hook, the node is created per normal. If a developer implements a hook, they can have some additional code run whenever a node is created. This code could do anything, including rolling back and/or altering the original action. It could also do something unrelated to the node creation entirely.
A callback could be thought of as a specific kind of hook. By implementing callback functionality into a system, that system is letting you call some additional code after an action has completed. However, hooking (as a generic term) is not limited to callbacks.
Another example. Sometimes Web Developers will refer to class names and/or IDs on elements as hooks. That's because by placing the ID/class name on an element, they can then use Javascript to modify that element, or "hook in" to the page document. (this is stretching the meaning, but it is commonly used and worth mentioning)
Oftentimes hooking refers to Win32 message hooking or the Linux/OSX equivalents, but more generically hooking is simply notifying another object/window/program/etc that you want to be notified when a specified action happens. For instance: Having all windows on the system notify you as they are about to close.
As a general rule, hooking is somewhat hazardous since doing it without understanding how it affects the system can lead to instability or at the very leas unexpected behaviour. It can also be VERY useful in certain circumstances, thought. For instance: FRAPS uses it to determine which windows it should show it's FPS counter on.
A hook is functionality provided by software for users of that software to have their own code called under certain circumstances. That code can augment or replace the current code.
In the olden days when computers were truly personal and viruses were less prevalent (I'm talking the '80's), it was as simple as patching the operating system software itself to call your code. I remember writing an extension to the Applesoft BASIC language on the Apple II which simply hooked my code into the BASIC interpreter by injecting a call to my code before any of the line was processed.
Some computers had pre-designed hooks, one example being the I/O stream on the Apple II. It used such a hook to inject the whole disk sub-system (Apple II ROMs were originally built in the days where cassettes were the primary storage medium for PCs). You controlled the disks by printing the ASCII code 4 (CTRL-D) followed by the command you wanted to execute then a CR, and it was intercepted by the disk sub-system, which had hooked itself into the Apple ROM print routines.
So for example, the lines:
PRINT CHR(4);"CATALOG"
PRINT CHR(4);"IN#6"
would list the disk contents then re-initialize the machine. This allowed such tricks as protecting your BASIC programs by setting the first line as:
123 REM XIN#6
then using POKE to insert the CTRL-D character in where the X was. Then, anyone trying to list your source would send the re-initialize sequence through the output routines where the disk sub-system would detect it.
That's often the sort of trickery we had to resort to, to get the behavior we wanted.
Nowadays, with the operating system more secure, it provides facilities for hooks itself, since you're no longer supposed to modify the operating system "in-flight" or on the disk.
They've been around for a long time. Mainframes had them (called exits) and a great deal of mainframe software uses those facilities even now. For example, the free source code control system that comes with z/OS (called SCLM) allows you to entirely replace the security subsystem by simply placing your own code in the exit.
In a generic sense, a "hook" is something that will let you, a programmer, view and/or interact with and/or change something that's already going on in a system/program.
For example, the Drupal CMS provides developers with hooks that let them take additional action after a "content node" is created. If a developer doesn't implement a hook, the node is created per normal. If a developer implements a hook, they can have some additional code run whenever a node is created. This code could do anything, including rolling back and/or altering the original action. It could also do something unrelated to the node creation entirely.
A callback could be thought of as a specific kind of hook. By implementing callback functionality into a system, that system is letting you call some additional code after an action has completed. However, hooking (as a generic term) is not limited to callbacks.
Another example. Sometimes Web Developers will refer to class names and/or IDs on elements as hooks. That's because by placing the ID/class name on an element, they can then use Javascript to modify that element, or "hook in" to the page document. (this is stretching the meaning, but it is commonly used and worth mentioning)
Simple said:
A hook is a means of executing custom code (function) either before, after, or instead of existing code. For example, a function may be written to "hook" into the login process in order to execute a Captcha function before continuing on to the normal login process.
A chain of hooks is a set of functions in which each function calls the next. What is significant about a chain of hooks is that a programmer can add another function to the chain at run time. One way to do this is to look for a known location where the address of the first function in a chain is kept. You then save the value of that function pointer and overwrite the value at the initial address with the address of the function you wish to insert into the hook chain. The function then gets called, does its business and calls the next function in the chain (unless you decide otherwise). Naturally, there are a number of other ways to create a chain of hooks, from writing directly to memory to using the metaprogramming facilities of languages like Ruby or Python.
An example of a chain of hooks is the way that an MS Windows application processes messages. Each function in the processing chain either processes a message or sends it to the next function in the chain.
Source: Stackoverflow.com