Architecture Questions

Here is an overview of current questions prepared for easier review of architecture of modules written on top of NetBeans platform. The questions should guide everyone thru a self review of its own module and help people recognize various aspects of API development and prevent various kinds of mistakes leading to easier maintanence and also for a base for documentation of the project as the answers form useful part of javadoc documentation.

Before Starting Work

Here is a set of basic questions one should answer when considering to start a work on new API exposing module or any other general functionality. To generate the answers to these questions please follow general instructions to create the XML answers and Ant script file and then execute
  ant -Darch.when=init
to generate the HTML page containing answers to these questions.

Question (arch-what): What is this project good for?
Hint: Please provide here a few lines describing the project, what problem it should solve, provide links to documentation, specifications, etc.

Question (arch-overall): Describe the overall architecture.
Hint: What will be API for clients and what support API? What parts will be pluggable? How will plug-ins be registered? Please use <api type="export"/> to describe your general APIs and specify their stability categories. If possible please provide simple diagrams.

Question (arch-usecases):
Hint: Content of this answer will be displayed as part of page at http://www.netbeans.org/download/dev/javadoc/usecases.html You can use tags <usecase name="name> regular html description </usecase> and if you want to use an URL you can prefix if with @TOP@ to begin at the root of your javadoc Describe the main use cases of the new API. Who will use it under what circumstances? What kind of code would typically need to be written to use the module?

Question (arch-time): What are the time estimates of the work?
Hint: Please express your estimates of how long the design, implementation, stabilization are likely to last. How many people will be needed to implement this and what is the expected milestone by which the work should be ready?

Question (arch-quality): How will the quality of your code be tested and how are future regressions going to be prevented?
Hint: What kind of testing do you want to use? How much functionality, in which areas, should be covered by the tests? How you find out that your project was successful?

Question (dep-nb): What other NetBeans projects and modules does this one depend on?
Hint: Depending on other NetBeans projects influnces the ability of users of your work to customize their own branded version of NetBeans by enabling and disabling some modules. Too much dependencies restrict this kind of customization. If that is your case, then you may want to split your functionality into pieces of autoload, eager and regular modules which can be enabled independently. Usually the answer to this question is generated from your project.xml file, but if it is not guessed correctly, you can suppress it by specifying <defaultanswer generate="none"/> and write here your own. Please describe such projects as imported APIs using the <api name="identification" type="import or export" category="stable" url="where is the description" />. By doing this information gets listed in the summary page of your javadoc.

Question (dep-non-nb): What other projects outside NetBeans does this one depend on?
Hint: Depending on 3rd party libraries is always problematic, especially if they are not open source, as that complicates the licensing scheme of NetBeans. Please enumerate your external dependencies here, so it is correctly understood since the begining what are the legal implications of your project. Also please note that some non-NetBeans projects are packaged as NetBeans modules (see libraries) and it is preferred to use this approach when more modules may depend and share such third-party libraries.

Question (dep-platform): On which platforms does your module run? Does it run in the same way on each?
Hint: If you plan any dependency on OS or any usage of native code, please describe why you are doing so and describe how you envision to enforce the portability of your code. Please note that there is a support for OS conditionally enabled modules which together with autoload/eager modules can allow you to enable to provide the best OS aware support on certain OSes while providing compatibility bridge on the not supported ones. Also please list the supported OSes/HW platforms and mentioned the lovest version of JDK required for your project to run on. Also state whether JRE is enough or you really need JDK.

Question (deploy-packages): Are packages of your module made inaccessible by not declaring them public?
Hint: By default NetBeans build harness treats all packages are private. If you export some of them - either as public or friend packages, you should have a reason. If the reason is described elsewhere in this document, you can ignore this question.

Question (compat-standards): Does the module implement or define any standards? Is the implementation exact or does it deviate somehow?

Question (compat-deprecation): How the introduction of your project influences functionality provided by previous version of the product?
Hint: If you are planning to deprecate/remove/change any existing APIs, list them here accompanied with the reason explaining why you are doing so.

Question (lookup-lookup): Does your module use org.openide.util.Lookup or any similar technology to find any components to communicate with? Which ones?
Hint: NetBeans is build around a generic registry of services called lookup. It is preferable to use it for registration and discovery if possible. See The Solution to Comunication Between Components . If you do not plan to use lookup and insist usage of other solution, then please describe why it is not working for you.
When filling the final version of your arch document, please describe the interfaces you are searching for, where are defined, whether you are searching for just one or more of them, if the order is important, etc. Also classify the stability of such API contract. Use <api group=&lookup& /> tag, so your information gets listed in the summary page of your javadoc.

Question (exec-threading): What threading models, if any, does your module adhere to? How the project behaves with respect to threading?
Hint: Is your API threadsafe? Can it be accessed from any threads or just from some dedicated ones? Any special relation to AWT and its Event Dispatch thread? Also if your module calls foreign APIs which have a specific threading model, indicate how you comply with the requirements for multithreaded access (synchronization, mutexes, etc.) applicable to those APIs. If your module defines any APIs, or has complex internal structures that might be used from multiple threads, declare how you protect data against concurrent access, race conditions, deadlocks, etc., and whether such rules are enforced by runtime warnings, errors, assertions, etc. Examples: a class might be non-thread-safe (like Java Collections); might be fully thread-safe (internal locking); might require access through a mutex (and may or may not automatically acquire that mutex on behalf of a client method); might be able to run only in the event queue; etc. Also describe when any events are fired: synchronously, asynchronously, etc. Ideas: Threading Recommendations (in progress)

Question (perf-scale): Which external criteria influence the performance of your program (size of file in editor, number of files in menu, in source directory, etc.) and how well your code scales?
Hint: Please include some estimates, there are other more detailed questions to answer in later phases of implementation.

Question (perf-limit): Are there any hard-coded or practical limits in the number or size of elements your code can handle?
Hint: Most of algorithms have increasing memory and speed complexity with respect to size of data they operate on. What is the critical part of your project that can be seen as a bottleneck with respect to speed or required memory? What are the practical sizes of data you tested your project with? What is your estimate of potential size of data that would cause visible performance problems? Is there some kind of check to detect such situation and prevent "hard" crashes - for example the CloneableEditorSupport checks for size of a file to be opened in editor and if it is larger than 1Mb it shows a dialog giving the user the right to decide - e.g. to cancel or commit suicide.

Question (perf-spi): How the performance of the plugged in code will be enforced?
Hint: If you allow foreign code to be plugged into your own module, how do you enforce that it will behave correctly and quickly and will not negatively influence the performance of your own module?

During implementation

When designing the implemenation of the project please keep following questions in mind. If you want to see the generated HTML answers use
  ant -Darch.when='init impl'
and both sets of questions, the initial one and this one will be generated.

Question (arch-where): Where one can find sources for your module?
Hint: Please provide link to the CVS web client at http://www.netbeans.org/download/source_browse.html or just use tag defaultanswer generate='here'

Question (deploy-jar): Do you deploy just module JAR file(s) or other files as well?
Hint: Usually a module consist of one JAR file (perhaps with Class-Path extensions) and also a configuration file that enables it. If you have any other files, use <api group="java.io.File" name="yourname" type="export" category="friend">...</api> to define the location, name and stability of your files (of course changing "yourname" and "friend" to suit your needs). If it uses more than one JAR, describe where they are located, how they refer to each other. If it consist of module JAR(s) and other files, please describe what is their purpose, why other files are necessary. Please make sure that installation/uninstallation leaves the system in state as it was before installation.

Question (deploy-nbm): Can you deploy an NBM via the Update Center?
Hint: If not why?

Question (compat-i18n): Is your module correctly internationalized?
Hint: Correct internationalization means that it obeys instructions at NetBeans I18N pages.

Question (compat-version): Can your module coexist with earlier and future versions of itself? Can you correctly read all old settings? Will future versions be able to read your current settings? Can you read or politely ignore settings stored by a future version?
Hint: Very helpful for reading settings is to store version number there, so future versions can decide whether how to read/convert the settings and older versions can ignore the new ones.

Question (exec-property): Is execution of your code influenced by any environment or Java system (System.getProperty) property? On a similar note, is there something interesting that you pass to java.util.logging.Logger? Or do you observe what others log?
Hint: If there is a property that can change the behavior of your code, somebody will likely use it. You should describe what it does and the stability category of this API. You may use
                <api type="export" group="property" name="id" category="private" url="http://...">
                    description of the property, where it is used, what it influence, etc.
                </api>            
            

Question (exec-component): Is execution of your code influenced by any (string) property of any of your components?
Hint: Often JComponent.getClientProperty, Action.getValue or PropertyDescriptor.getValue, etc. are used to influence a behavior of some code. This of course forms an interface that should be documented. Also if one depends on some interface that an object implements (component instanceof Runnable) that forms an API as well.

Question (exec-ant-tasks): Do you define or register any ant tasks that other can use?
Hint: If you provide an ant task that users can use, you need to be very careful about its syntax and behaviour, as it most likely forms an API for end users and as there is a lot of end users, their reaction when such API gets broken can be pretty strong.

Question (exec-classloader): Does your code create its own class loader(s)?
Hint: A bit unusual. Please explain why and what for.

Question (exec-reflection): Does your code use Java Reflection to execute other code?
Hint: This usually indicates a missing or insufficient API in the other part of the system. If the other side is not aware of your dependency this contract can be easily broken.

Question (exec-process): Do you execute an external process from your module? How do you ensure that the result is the same on different platforms? Do you parse output? Do you depend on result code?
Hint: If you feed an input, parse the output please declare that as an API.

Question (exec-introspection): Does your module use any kind of runtime type information (instanceof, work with java.lang.Class, etc.)?
Hint: Check for cases when you have an object of type A and you also expect it to (possibly) be of type B and do some special action. That should be documented. The same applies on operations in meta-level (Class.isInstance(...), Class.isAssignableFrom(...), etc.).

Question (format-types): Which protocols and file formats (if any) does your module read or write on disk, or transmit or receive over the network? Do you generate an ant build script? Can it be edited and modified?
Hint:

Files can be read and written by other programs, modules and users. If they influence your behaviour, make sure you either document the format or claim that it is a private api (using the <api> tag).

If you generate an ant build file, this is very likely going to be seen by end users and they will be attempted to edit it. You should be ready for that and provide here a link to documentation that you have for such purposes and also describe how you are going to understand such files during next release, when you (very likely) slightly change the format.

Question (format-dnd): Which protocols (if any) does your code understand during Drag & Drop?
Hint: Often Node's deal with clipboard by usage of Node.drag, Node.getDropType. Check your code for overriding these methods. Btw. if they are not overridden, they by default delegate to Node.clipboardCopy, Node.clipboardCut and Node.pasteTypes.

Question (format-clipboard): Which data flavors (if any) does your code read from or insert to the clipboard (by access to clipboard on means calling methods on java.awt.datatransfer.Transferable?
Hint: Often Node's deal with clipboard by usage of Node.clipboardCopy, Node.clipboardCut and Node.pasteTypes. Check your code for overriding these methods.

Before Commit

Before claiming that the module have reached final state and is ready for use, go thru all the previous questions and verify that they are still valid and also check the following set of questions.

Question (dep-jre): Which version of JRE do you need (1.2, 1.3, 1.4, etc.)?
Hint: It is expected that if your module runs on 1.x that it will run on 1.x+1 if no, state that please. Also describe here cases where you run different code on different versions of JRE and why.

Question (dep-jrejdk): Do you require the JDK or is the JRE enough?

Question (deploy-shared): Do you need to be installed in the shared location only, or in the user directory only, or can your module be installed anywhere?
Hint: Installation location shall not matter, if it does explain why. Consider also whether InstalledFileLocator can help.

Question (deploy-dependencies): What do other modules need to do to declare a dependency on this one, in addition to or instead of the normal module dependency declaration (e.g. tokens to require)?
Hint: Provide a sample of the actual lines you would add to a module manifest to declare a dependency, for example OpenIDE-Module-Requires: some.token. If other modules should not depend on this module, or should just use a simple regular module dependency, you can just answer "nothing". If you intentionally expose a semistable API to clients using implementation dependencies, you should mention that here (but there is no need to give an example of usage).

Question (resources-file): Does your module use java.io.File directly?
Hint: NetBeans provide a logical wrapper over plain files called org.openide.filesystems.FileObject that provides uniform access to such resources and is the preferred way that should be used. But of course there can be situations when this is not suitable.

Question (resources-layer): Does your module provide own layer? Does it create any files or folders in it? What it is trying to communicate by that and with which components?
Hint: NetBeans allows automatic and declarative installation of resources by module layers. Module register files into appropriate places and other components use that information to perform their task (build menu, toolbar, window layout, list of templates, set of options, etc.).

Question (resources-read): Does your module read any resources from layers? For what purpose?
Hint: As this is some kind of intermodule dependency, it is a kind of API. Please describe it and classify according to common stability categories.

Question (resources-mask): Does your module mask/hide/override any resources provided by other modules in their layers?
Hint: If you mask a file provided by another module, you probably depend on that and do not want the other module to (for example) change the file's name. That module shall thus make that file available as an API of some stability category.

Question (resources-preferences): Does your module uses preferences via Preferences API? Does your module use NbPreferences or or regular JDK Preferences ? Does it read, write or both ? Does it share preferences with other modules ? If so, then why ?
Hint: You may use <api type="export" group="preferences" name="preference node name" category="private"> description of individual keys, where it is used, what it influences, whether the module reads/write it, etc. </api> Due to XML ID restrictions, rather than /org/netbeans/modules/foo give the "name" as org.netbeans.modules.foo. Note that if you use NbPreferences this name will then be the same as the code name base of the module.

Question (lookup-register): Do you register anything into lookup for other code to find?
Hint: Do you register using layer file or using META-INF/services? Who is supposed to find your component?

Question (lookup-remove): Do you remove entries of other modules from lookup?
Hint: Why? Of course, that is possible, but it can be dangerous. Is the module your are masking resource from aware of what you are doing?

Question (exec-privateaccess): Are you aware of any other parts of the system calling some of your methods by reflection?
Hint: If so, describe the "contract" as an API. Likely private or friend one, but still API and consider rewrite of it.

Question (security-policy): Does your functionality require modifications to the standard policy file?
Hint: Your code might pass control to third-party code not coming from trusted domains. This could be code downloaded over the network or code coming from libraries that are not bundled with NetBeans. Which permissions need to be granted to which domains?

Question (security-grant): Does your code grant additional rights to some other code?
Hint: Avoid using a class loader that adds extra permissions to loaded code unless really necessary. Also note that your API implementation can also expose unneeded permissions to enemy code by calling AccessController.doPrivileged().

Question (perf-startup): Does your module run any code on startup?

Question (perf-exit): Does your module run any code on exit?

Question (perf-mem): How much memory does your component consume? Estimate with a relation to the number of windows, etc.

Question (perf-wakeup): Does any piece of your code wake up periodically and do something even when the system is otherwise idle (no user interaction)?

Question (perf-progress): Does your module execute any long-running tasks?
Hint: Long running tasks should never block AWT thread as it badly hurts the UI responsiveness. Tasks like connecting over network, computing huge amount of data, compilation be done asynchronously (for example using RequestProcessor), definitively it should not block AWT thread.

Question (perf-huge_dialogs): Does your module contain any dialogs or wizards with a large number of GUI controls such as combo boxes, lists, trees, or text areas?

Question (perf-menus): Does your module use dynamically updated context menus, or context-sensitive actions with complicated and slow enablement logic?
Hint: If you do a lot of tricks when adding actions to regular or context menus, you can significantly slow down display of the menu, even when the user is not using your action. Pay attention to actions you add to the main menu bar, and to context menus of foreign nodes or components. If the action is conditionally enabled, or changes its display dynamically, you need to check the impact on performance. In some cases it may be more appropriate to make a simple action that is always enabled but does more detailed checks in a dialog if it is actually run.

Project Features

About this Project

openide was started in November 2009, is owned by Antonin Nebuzelsky, and has 85 members.
By use of this website, you agree to the NetBeans Policies and Terms of Use (revision 20160708.bf2ac18). © 2014, Oracle Corporation and/or its affiliates. Sponsored by Oracle logo
 
 
Close
loading
Please Confirm
Close