Read/write lock

In general, code-related data structures in the IntelliJ Platform are covered by a single reader/writer lock. This applies to PSI, VFS and project root model.

Reading data is allowed from any thread. Reading data from the UI thread does not require any special effort. However, read operations performed from any other thread need to be wrapped in a read action by using ApplicationManager.getApplication().runReadAction() or, shorter, ReadAction.run/compute.

Writing data is only allowed from the UI thread, and write operations always need to be wrapped in a write action with ApplicationManager.getApplication().runWriteAction() or, shorter, WriteAction.run/compute.

In addition, modifying the model is only allowed from write-safe contexts, which include user actions, invokeLater calls from them (see the next section), and transactions (TransactionGuard.submitTransaction). You may not modify PSI, VFS or project model from inside UI renderers or SwingUtilities.invokeLater calls. See TransactionGuard documentation for more details.

You must not access the model outside a read or write action. The corresponding objects are not guaranteed to survive between several consecutive read actions. So as a rule of thumb, whenever you start a read action, you should check if your PSI/VFS/project/module are still valid.

invokeLater

To pass control from a background thread to the event dispatch thread, instead of the standard SwingUtilities.invokeLater(), plugins should use ApplicationManager.getApplication().invokeLater(). The latter API allows specifying the modality state for the call, i.e. the stack of modal dialogs under which the call is allowed to execute.

  • Passing ModalityState.NON_MODAL means that the operation will be executed after all modal dialogs are closed. Note that if any of the open (unrelated) project displays a per-project modal dialog, the action will be executed after the dialog is closed.
  • Passing ModalityState.stateForComponent() means that the operation can be executed when the topmost shown dialog is the one that contains the specified component, or is one of its parent dialogs.
  • If no modality state is passed, ModalityState.defaultModalityState() will be used. In most cases, this is the optimal choice, that uses current modality state when invoked from UI thread, and has a special handling for background processes started with ProgressManager: invokeLater from such a process may run in the same dialog that the process started.
  • ModalityState.any() means that the runnable will be executed as soon as possible regardless of modal dialogs. Please note that modifying PSI, VFS or project model is prohibited from such runnables. See TransactionGuard documentation for more details.

If your UI thread activity needs to access file-based index (e.g. it's doing any kind of project-wide PSI analysis, resolves references, etc), please use DumbService#smartInvokeLater. That way, your activity will be run after all possible indexing processes have been completed.

Background processes and ProcessCanceledException

Background progresses are managed by ProgressManager class, which has plenty of methods to execute the given code with a modal (dialog), non-modal (visible in the status bar) or invisible progress. In all cases, the code is executed on a background thread which is associated with a ProgressIndicator object. The current thread's indicator can be retrieved any time via ProgressIndicatorProvider#getGlobalProgressIndicator.

For visible progresses, threads can use ProgressIndicator to notify the user about current status: e.g. set text or visible fraction of the work done.

Progress indicators also provide means to handle cancellation of background processes, either by user (pressing "Cancel" button), or from code (e.g. when the current operation becomes obsolete due to some changes in the project). The progress can be marked as canceled by calling ProgressIndicator#cancel. The process reacts to this by calling ProgressIndicator#checkCanceled (or ProgressManager#checkCanceled if you don't have indicator instance at hand). This call throws a special unchecked ProcessCanceledException if the background process has been canceled.

All code working with PSI, or in other kinds of background processes, should be prepared to a ProcessCanceledException being thrown from any point. This exception should never be logged, it should be rethrown, and it'll be handled in the infrastructure that started the process.

The checkCanceled should be called often enough to guarantee smooth cancellation of the process. PSI internals have a lot of checkCanceled calls inside. But if your process does lengthy non-PSI activity, you might need to insert explicit checkCanceled calls so that it happens frequently, e.g. on each Nth loop iteration.

Preventing UI freezes

Background threads shouldn't take read actions for a long time. The reason is that if the UI thread needs a write action (e.g. the user types something), it must acquire it as soon as possible, otherwise the UI will freeze until all background threads have released their read actions.

The best known approach to that is to cancel background read actions whenever there's a write action about to occur, and restart that background read action later from the scratch. Editor highlighting, code completion, Goto Class/File/etc actions all work like this. To achieve that, the lengthy background operation is started with a ProgressIndicator, and a special listener cancels that indicator when write action is started. The next time the background thread calls checkCanceled, a ProcessCanceledException will be thrown, and the thread should stop its operation (and finish the read action) as soon as possible.

There are two recommended ways of doing this:

  • If you're on UI thread, create a ReadTask and pass it to one of ProgressIndicatorUtils.schedule* methods. Inside onCanceled method, schedule it again if the activity should be restarted.
  • If you're already in a background thread, use ProgressManager.getInstance().runInReadActionWithWriteActionPriority() in a loop, until it passes or the whole activity becomes obsolete.

In both approaches, you should always check at the start of each read action, if the objects you're working with are still valid, and if the whole operation still makes sense (i.e. not canceled by the user, the project isn't closed, etc.)

If the activity you're doing has to access file-based index (e.g. it's doing any kind of project-wide PSI analysis, resolves references, etc), you should override ReadTask#runBackgroundProcess and use "smart-mode" read action there: DumbService.getInstance(project).runReadActionInSmartMode(() -> performInReadAction(indicator))

results matching ""

    No results matching ""