1 / 61

Plug-in Architectures

Plug-in Architectures. J effrey S tylos User Interface Software, Fall 2004. Introduction. Component architectures. What are components good for? Being able to reuse standard components within your own applications Allowing applications to be architected and implemented more modularly

tasmine
Download Presentation

Plug-in Architectures

An Image/Link below is provided (as is) to download presentation Download Policy: Content on the Website is provided to you AS IS for your information and personal use and may not be sold / licensed / shared on other websites without getting consent from its author. Content is provided to you AS IS for your information and personal use only. Download presentation by click this link. While downloading, if for some reason you are not able to download a presentation, the publisher may have deleted the file from their server. During download, if you can't get a presentation, the file might be deleted by the publisher.

E N D

Presentation Transcript


  1. Plug-in Architectures Jeffrey Stylos User Interface Software, Fall 2004

  2. Introduction

  3. Component architectures • What are components good for? • Being able to reuse standard components within your own applications • Allowing applications to be architected and implemented more modularly • Allowing applications to be extended (by plug-ins) after they are compiled and released

  4. What is a plug-in? • For my purposes, a plug-in is a piece of software that extends or changes the behavior or interface of an existing, compiled application

  5. What are plug-ins good for? • Extending an application’s functionality • Changing the UI of an application (skins or themes) • Extracting data from an application • To connect it to another app • For logging

  6. What are the issues with plug-in architectures? • What sort of plug-ins can you create? • What aspects of the original program can you override? • How difficult does this make it to create the base application? • How are the plug-ins created and installed? Are they compiled? • How do plug-in authors debug their plug-ins? • How you create an architecture that balance these different issues is an interesting research question

  7. My goals for this lecture • Give an overview of several different examples of plug-in architectures and their capabilities and limitations • Give you an idea of what resources and tools to use if you want to create one of these plug-ins • Not to die from coughing

  8. Why you should care about plug-in architectures • Writing plug-ins is a way to connect your research with real applications, making it more compelling • Writing plug-ins is a relatively easy way to write software that can make your life easier • Building a plug-in architecture into your project is a way to let other people improve your project for you • Designing new plug-in architectures is a potentially fruitful research avenue

  9. DISCLAIMER • This wasn’t one of the existing topics • I picked it because I wanted to learn more, not because I was already an expert • These slides suck • I have a cough

  10. Overview • Intro • Emacs • Mozilla • Eclipse • Skinnable apps

  11. Emacs

  12. Emacs • "Editor MACroS“ • One of the first runtime extensible systems • Designed for text editing • Used for everything • Contains a core written in C • The rest is written in elisp

  13. Emacs extensions • Extensions are written in elisp • Can provide new functions or replace existing functionality • Most extensions are macros that provide some sort of automation when explicitly invoked • Can also do things like change text formatting • Can display images, but only in the context of lines of text • Auto spell-checking (flyspell) • It’s hard to: • Write multi-threaded extensions (makes network programming difficult) • Do arbitrary painting (assumes lines of text)

  14. Emacs extensions (continued) • Everything loaded from initialization file • To install an extension • Set the correct paths in the initialization file • Load the extension from the initialization file • Can also manually load any lisp file, or even execute code as you write it

  15. How do you debug extensions? • Because much of Emacs is in Lisp, can step through it • Has a built-in debugging mode • Extensions for more detailed debugging (Edebug) and profiling (ELP) • Don't have to build Emacs to run or debug an extension

  16. Emacs examples • [Show Tetris example]

  17. Mozilla

  18. Mozilla • Several different applications • Current code base provides • Mozilla 1.8 • Netscape 7.2 • Firefox 1.0 • Thunderbird 1.0 • Different from old Netscape (which also provided some extensibility) • Current extensibility features allow modification of the browser interface components (not just interactive webpage elements)

  19. Mozilla extensibility architectures • XPCOM • Heavier weight – supports C++ and other programming languages • XUL / JavaScript • User interface modification and scripting • Also used for “themes” • [XPConnect] • Bridges XPCOM and Javascript • [XPInstall] • Installation and package management

  20. XPCOM • Cross-Platform COM • A cross platform clone of COM • Multi-threaded, but no remote support (not DCOM) • Used to create heavy-weight components that can be called by XUL and JavaScript elements • Code can be cross platform, but has to be recompiled

  21. XPCOM (continued) • Used to implement much of Mozilla’s core functionality

  22. XPCOM (continued) • Uses the IDL (Interface Definition Language) to specify capabilities • IDL compiles to C++ • Connects to JavaScript • As with COM, XPCOM programming uses lots of interface pointers • §Bad:void ProcessSample(nsSampleImpl* aSample) {    aSample->Poke(“hello”);}§Good:void ProcessSample(nsISample* aSample) {    aSample->Poke(“Hello”);}

  23. Using XPCOM • (Usually) needs Mozilla framework to compile • Can be difficult to compile • Needs the Mozilla framework to debug component • Requires special tool to generate IDL specification • Can use XPCOM for programming other than Mozilla plug-ins, but few do

  24. Debugging XPCOM components • Effective when you debug all of Mozilla • Tools for detecting memory leaks • BloatView, Boehm GC Leak Detector, Refcount Tracing, Trace-Malloc, etc.

  25. High level overview of XPCOM Mozilla plug-ins • Make a DLL that implements necessary methods (NSGetModule, etc.) • Create a IDL specification • Create an XPT specification from the IDL • Put everything in Mozilla’s components directory

  26. XPCOM Examples • Using XPCOM: • var sound = Components.classes["@mozilla.org/sound;1"].createInstance(); • if (sound) sound.QueryInterface(Components.interfaces.nsISound); • IDL: • [scriptable, uuid(1BDC2EE0-E92A-11d4-BCC0-0060089296CB)] • interface mozIGPS : nsISupports • { • boolean Open(in string strDevice); • boolean Close(); • string Reason(in boolean bClear); • readonly attribute double latitude; • readonly attribute double longitude; • readonly attribute double elevation; • readonly attribute double gpstime; • };

  27. XPCOM resources • High level information • http://www.mozilla.org/projects/xpcom/ • An Introduction to XPCOM • http://www-128.ibm.com/developerworks/webservices/library/co-xpcom.html • I don’t know of any good tutorial for creating and compiling an XPCOM extension

  28. Mozilla Plug-ins: XUL and JavaScript • Used to build the base Mozilla user interface • What most of the available extensions are written in • Uses XUL to specify the static layout of the UI elements • Uses JavaScript to specify how each of the elements interacts

  29. XUL • “zool” • XML User-interface Language • Initial ambitions of also being a language for richer webpages (sort of works now) • Cross platform • Can be used to create stand alone applications (with difficulty)

  30. XUL (continued) • Uses “overlays” to add new interface elements to existing interfaces • Easy to add a new button to an existing form • Can also replace existing interface elements • Uses CSS to specify many of the formatting elements of the UI

  31. How to write a XUL-based plug-in • Write a .xul and .js file in Notepad • Put in Mozilla’s extension directory

  32. Debugging XUL and JavaScript plug-ins • Difficult – most common solution is trial and error • Prints out red error messages on the UI • There exists a debugging tool • Venkman JavaScript debugger • …but it hasn’t worked with Firefox or Thunderbird for a year or so • Update: There’s a development branch that might work with some version of Firefox

  33. Example XUL code • <?xml version="1.0"?> • <!DOCTYPE window> • <?xul-overlay href="file://src/HTML/moztests/sampleoverlay.xul"?> • <window orient="vertical" • xmlns="http://www.mozilla.org/keymaster/gatekeeper/there.is.only.xul"> • <box orient="vertical"> • <box orient="horizontal" flex="1"> • <text value="Sample text number ONE"/> • <spring flex="1"/> • <text value="Sample text number TWO"/> • </box> • <box orient="vertical"> • <text value="Sample text number THREE"/> • <text value="Sample text number FOUR"/> • </box> • <box id="extraStuff"/> • </box> • </window>

  34. Example JavaScript / XUL code • function toggleContactsPicker() { • var sidebarBox = top.document.getElementById("sidebar-box"); • var sidebarSplitter = top.document.getElementById("contacts-pane-splitter"); • var sidebarState = sidebarSplitter.getAttribute("state"); • var menuItem = top.document.getElementById("menu_showContacts"); • var toolbarButton = top.document.getElementById("button-contacts"); • if (sidebarBox.hidden) { • sidebarBox.hidden = false; • sidebarSplitter.hidden = false; • if (menuItem) • menuItem.setAttribute("checked","true"); • if (toolbarButton) • toolbarButton.setAttribute("checked","true"); • ...

  35. Example XUL extensions • [Show Firefox and Thunderbird extensions] • [Thunderbird extensions: • Contact sidebar • Quote colors]

  36. Themes • Uses the swap-ability of XUL elements to create plug-ins that change the appearance but not the functionality • [Show theme examples]

  37. XUL and JavaScript resources • Lots, high-level overviews, reference sites, and examples • Because they are non-compiled languages, you can view the source of every current available extension

  38. XUL / JavaScript overview • Powerful and flexible UI scripting • Can override existing elements • Can dynamically transform UI elements • Provides support for separating formatting decisions into CSS files • Provides support for internationalization by separating text strings into language files • Don’t have to compile • Hard to debug • Can’t write arbitrary C code • Potentially slow

  39. Break

  40. Eclipse

  41. Eclipse • A modern Emacs • Designed to have a very small core that loads independent components • Written in Java, primary selling point is its included Java developing components • “The Eclipse Platform is an IDE for anything, and for nothing in particular.” • Cross platform

  42. Eclipse (continued) • Designed for building IDEs • People used it for other apps • Remail, Haystack (new version) • Now formalized: “Rich Client Protocol”

  43. Eclipse architecture

  44. Eclipse plug-ins • All written in Java • Found at Eclipse launch • Heavier-weight than Emacs model – can’t dynamically swap plug-ins or write and execute code • Include a “manifest” file • Specifies visibility of included classes and methods • Used to add information to the plug-in registry

  45. Manifest file example • <?xml version="1.0" encoding="UTF-8"?> • <plugin • name="JUnit Testing Framework" • id="org.junit" • version="3.7" • provider-name="Eclipse.org"> • <runtime> • <library name="junit.jar"> • <export name="*"/> • </library> • </runtime> • </plugin>

  46. Eclipse plug-in architecture • Load-on-demand strategy makes it feasible to have many different plug-ins and still have reasonable performance • “Extension points” make it easy for plug-ins to themselves be extendable • Allows for multi-level extensibility • (Most architectures only support a single level of extensibility) • Uses “explicit ports” to make plug-in connections clear • Plug-ins say what they can extend • Helps support multi-layered extensibility • Manifests help encapsulate plug-ins from each other (they can only use each other in specified ways) • Again, helps multi-layered extensibility

  47. Eclipse plug-in architecture (cont.) • Limitations: • Extensibility points are tied to specific implementations • Can’t have multiple swappable implementations of the same functionality • Can’t have strict dependencies • All components are optional • Can’t say “this plug-in only works when this other plug-in is available” • Reference: Evaluating the Eclipse Platform as a Composition Environment

  48. SWT and JFace • Standard Widget Toolkit • Really just a widget set • JFace is the higher-level toolkit built on top of SWT • Allows for portable plug-ins that use native widgets • Usable outside of Eclipse • Opinionated claim: Best Java widget set • Supports widget extensibility by allowing Java extensions of native widgets

  49. Different types of Eclipse plug-ins • Whole apps • C++ IDE (CDE) • Visual UI Editor (SWT/Swing Designer) • Aspect oriented language extension (AspectJ) • Profiler (EclipseProfiler)

  50. Sample Eclipse plug-in • [Show HelloWorld Eclipse plug-in]

More Related