String Handling Plug-in

["Jonathan Gossage" <jgossage@xxxxxxxx>]

I have produced a first draft of the requirements for a tool to help with
the automated management of strings in program text where
internationalization is required. This document deliberately does not
address design issues at present. It simply deals with requirements.

I would like to get feedback if possible on the feelings of the team
regarding the scope and appropriateness of the proposed functionality.

Regards

Jonathan

Personal Email
jgossage@xxxxxxxx

Business Email
jonathan@xxxxxxxxxxxxxx

Title: String Handling Plug-in

String Handling Plug-in

 

 

Jonathan Gossage

 

Copyright Ó Jonathan Gossage, 2002

 

 

 

Overview

This document examines the requirements for managing strings in an international development environment. Based on these requirements it then proposes an Eclipse based design that should significantly reduce the effort required to deal effectively with strings, particularly when internationalization is a requirement.

Why not use current Eclipse support

Eclipse already provides a tool to support string internationalization. What are the reasons for developing another one?  The reasons fall into two main areas:

Usability

On the Stellation sub-project of Eclipse, the Eclipse tool was used as an experiment on one package and the results gave the developers significant problems. They found the resulting code much harder to read and understand and they felt that the effort required to create the string environment was too high.

They also found that the amount of context shown when strings were being batch internationalized was insufficient to always be able to make good decisions about the handling of individual strings.

Scope

The Eclipse package only addresses a part of the entire string management picture. In particular, it does not address the problem of string consolidation where it is desired to only have one string with a particular semantic meaning and content. It also provides no assistance during the language translation phase of string internationalization. Nor does it optimize resource usage at runtime.

Requirements

Scope

General

This tool is intended to help manage text strings embedded in program text. It is not intended to provide any form of general text management capabilities.

Single language target

The basic service that can be provided in this scenario is to permit identical strings to be referenced symbolically where the duplicate copies have the same semantic meaning rather than just the same textual value. This helps with program maintenance since a change to the text of the string need only be made once.

Multi-language target

Services will be provided to create, maintain and access language and cultural specific versions of text strings. Access facilities will be provided for both development/editing stages and for program run-time. In addition, specialized services will be provided to aid in the translation process. These services will take full advantage of standard Java internationalization support.

Multi-part strings

Services will be provided to create and view multi-part strings that contain variable elements. Such strings pose particular problems during internationalization and these services will allow multi-part strings to be composed and will support the display of sample results while editing the strings. These services will take full advantage of Java support for message construction.

Usability

Entering strings into a program will simply require typing the string in most cases. The tool will take care of all the bookkeeping behind the scenes but will make the advanced capabilities available simply. Strings will be displayed on the screen as strings. The backing code needed to actually handle the string at run-time will be generated automatically and will only be displayed on demand. All generated code will be user editable when required.

Functionality

Convert existing conventional code bases

Users need to be able to convert strings in existing code bases into an internationalized and consolidated form. The specific requirements include the following:

• Identify the scope of the conversion effort. The conversion can be performed in one or more steps as required. The scope can be specified as an entire workspace, a set of projects or a set of files.
• Define the rules to be followed during the conversion. These include the default language, the location of the strings (i.e. the name and location of the string storage container), whether string translation will be required, whether duplicate strings will be automatically consolidated or whether strings should not be managed by default.

Convert existing Eclipse/Java managed code bases

A tool will be provided to convert environments that use variations on the basic Java strategy of using ResourceBundles, including code managed by the current Eclipse facilities. In addition the current Eclipse convention for marking strings that should not be tool-managed by adding a marker comment will be used and honored.

Editing the string

Strings will be entered, view and modified as normal text strings in the user’s normal language. The appearance will be of directly working on the string in the program text. Visual indicators will be provided for the string management rules in effect. These include string consolidation, internationalization and “don’t handle”. There may be a need for more than one method of indicating the current rules to ensure proper communication of state to the visually impaired.

Overrides for the current string management rules will be available via right click menus or via key chords that will be available while the cursor is within the bounds of the string.

A mechanism will be provided to allow the entry and display of long strings without the need for horizontal scrolling.

Advanced editing

Advanced editing comes into play when the user wants to display formatted text. The following java.text based formats will be supported.

• Number format
• Decimal format
• Choice format
• Message format (multi-part strings)

Both assisted and expert modes will be provided with auto-recognition of the mode chosen. This feature will include string preview that will show the result of the string formatting. The exact presentation of this feature will be subject to experimentation do determine the best human factors.

Editing the backing code

When string consolidation or internationalization is in effect, actual access to the string will be via code generated by the tool. Normally it should not be necessary to edit this code but the tool will provide a means to allow this if necessary. The same basic mechanism will also permit reading the backing code without editing when necessary.

Maintenance

Changes to the default language strings can be made at any time by a developer. When changes are made to any of the strings in a compilation unit that is internationalized, an Eclipse task will be generated to indicate that there is translation work to do for the compilation unit. Only one task per compilation unit will be outstanding for each target language for each compilation unit.

Configuration

A user can configure the following aspects of string management.

• Whether duplicate strings can be consolidated.
• Whether strings are to be internationalized.
• Which languages and cultural environments are to be supported for string internationalization? Languages and cultural environments may be added and removed at any time.
• The name of the resource file that will be used to store strings.

 

Configuration options can be specified on a workspace or project basis. Project options override workspace options.

Translation

The following services will be provided to support the translation of strings into the various target environments. A basic assumption will be that the translator will work in an Eclipse workspace, just as developers do.

• Eclipse tasks will be used to locate the units of work to be translated.
• Translators can specify the source and target languages/locales they will be working in. The default source language will be the default language for the workspace or project.
• A specialized translation editor will be provided that will provide the following services.
• Display of source and target strings.
• Display of string context.
• Editing of target strings.
• Evaluation of multi-part strings so the translator can see the result of translation decisions.
• The ability to generate Eclipse tasks identifying problem areas that must be worked on by a software developer before translation can continue. The ability to fully describe the problem will be included.

Extendibility

In keeping with the spirit of Eclipse, this tool will be designed as a framework with extension points that will permit the adaptation of the framework to different target environments. The initial implementation will target Java language development.

Deployment

Deployment can create special problems for string management. Individual classes and packages may be deployed into a multitude of environments and the same classes may be deployed into multiple environments at the same time. Tools are needed to ensure that al the associated strings are packaged and deployed with their associated classes.

Another problem is that the execution environment may vary considerably. Applets may be deployed into a security-constrained environment for example where there may be no access to the operating system environment. EJBs’ have their own set of constraints that must be met.

These considerations lead to the conclusion that the string data must be deployable in the same Jars as the application code that uses the strings.

Export

It is necessary to be able to move code using the mechanisms supported by this tool to other environments. Two facilities will be provided:

• Export the code with the strings from a user selected language/locale embedded in the source code.
• Export the code converted to use standard Java ResourceBundle techniques and property files.

Performance

String management is one of the areas that has been identified by the Eclipse team as a performance problem area. Specifically the techniques used by the Eclipse string internationalization tool have been cited (see http://www.eclipse.org/eclipse/development/performance/bloopers.html for details). This tool will use an optimized implementation for string storage and retrieval including lazy string retrieval to minimize the overhead in managing the string environment.

Design Overview