Difference between revisions of "Babel / Server Tool Development Process"
(→Obtain the code from CVS)
(→Set up the MySQL database)
|Line 63:||Line 63:|
=== Set up the MySQL database ===
=== Set up the MySQL database ===
Configure the /base.conf file with the MySQL database information.
Revision as of 09:55, 2 February 2009
- 1 Babel Team Development Strategy
- 2 Setting up a Development Environment
- 2.1 Install CVS, PHP, Apache, MySQL if needed
- 2.2 Install Java
- 2.3 Create a directory alias in Apache
- 2.4 Obtain the code from CVS
- 2.5 Write or reuse the addon
- 2.6 Set up the MySQL database
- 2.7 Requirements for PHP
- 2.8 Phoenix is no longer required for Babel Server
- 2.9 Logging in, load strings from dev.eclipse.org CVS
- 2.10 Creating (building) a language pack
- 2.11 Background jobs
- 2.12 Running the server on Windows
- 3 Submitting code patches
- 4 Committer Development Process
- 5 Submitting database changes
- 6 Release process
- 7 Data Bridges
Babel Team Development Strategy
The Babel team's strategy to processing bugs and feature requests is (in order):
- Bugs that cause incorrect translations to be saved or prevent people from enter translations
- Finishing the version zero issues (deployment, documentation)
- Bugs preventing project teams from adding their projects to Babel
- Server performance issues
- Bugs preventing users from consuming the output of Babel
- New features
Setting up a Development Environment
The Babel server environment is a typical LAMP environment. Apache 2.x, PHP 5.x, MySQL 5.x, Java and a CVS and/or SVN client are likely the only prerequisites.
Steps to setting up a Babel Development Environment on Linux. For Windows users, please refer to the Running the server on Windows section.
Install CVS, PHP, Apache, MySQL if needed
On Fedora 9, for example (as root):
# yum install mysql-server mysql php php-mysql httpd
You can choose to install cvs and/or svn depending on what repositories you will be working with:
# yum install cvs # yum install svn # /etc/init.d/mysqld start # /etc/init.d/httpd start
Install your favorite Java distro, and pay particular attention to have the jar command to be accessible in the path, since it will be used by the background jobs to create the language packs.
Create a directory alias in Apache
Edit (eg) /etc/httpd/conf/httpd.conf to add:
Alias /babel "/path/to/babel/server/html/" <Directory "/path/to/babel/server/html/"> AllowOverride None Order allow,deny Allow from all </Directory>
Apache may need to be reloaded or restarted for the alias to take effect. (eg Fedora 9: /etc/init.d/httpd restart)
Obtain the code from CVS
cd /path/to/babel/ cvs -d :pserver:email@example.com:/cvsroot/technology co -d server/ org.eclipse.babel/server
Write or reuse the addon
You need to select an addon in the addons folder to work with. You can either:
- Create a addons.conf file in the addons folder, and add this line in the file:
- Create an environment variable that will be used by the www-data user (or whatever user will be executing the php files):
If you want to develop using the Babel look and feel, you should be using the value "babel.eclipse.org". If you want to reuse the style of the staging instance, use "babel.eclipse.org/staging".
You can write your own addon, making it independent or reusing part of the existing addons.
Set up the MySQL database
Configure the addons/babel.eclipse.org/base.conf or addons/babel.eclipse.org/staging/base.conf (depending what you did above) file with the MySQL database information.
db_read_host=localhost db_read_user=babel db_read_pass=babelpassword db_read_name=babel context=dev
Using your favourite MySQL administration tool, create a database for Babel. It is also recommended to create a MySQL user for Babel.
mysql -u root -p mysql> create database babel; mysql> grant SELECT, INSERT, UPDATE, DELETE on babel.* to babel@'localhost' identified by 'babelpassword'; mysql> flush privileges;
Replace the usernames, hostname and password according to your specific environment. Load the babel tables and sample data (the root account is used here, as the 'babel' account doesn't have CREATE/DROP privileges):
mysql -u root -p babel < /path/to/babel/server/babel-setup.sql
The first time you'll run the Babel setup script, you will get an error from occurring for a trigger that does not exist.
To avoid this error, install MySQL version >= 5.0.32 or version >= 5.1.14 to be able to use IF EXISTS clause for DROP TRIGGER statement and modify the Babel sql script:
Replace: DROP TRIGGER `ins_version`;
by: DROP TRIGGER IF EXISTS `ins_version`;
Requirements for PHP
PHP must have the ability to encode data in the JSON format. As of PHP 5.2 'json_encode' and 'json_decode' are included by default. If you are using an earlier version you will need to install the JSON PECL extension, find another json_encode library or write your own json_encode function. See this page to find out more information on JSON and PHP.
Phoenix is no longer required for Babel Server
Previously, Eclipse Phoenix code was required for Babel Server to run. This requirement has been removed (bug 217488).
Logging in, load strings from dev.eclipse.org CVS
Log into Babel at http://localhost/babel/login.php.
username: firstname.lastname@example.org password: password
Launch the maps processor: http://localhost/babel/process_map_files.php
Creating (building) a language pack
To create a language pack, simply run php5 path/to/generate1.php from a command line (or from cron). This requires many Linux commands to be found in the PATH environment variable; notably, the jar command that is included with most JDKs.
The background jobs have evolved to a separate page.
Running the server on Windows
The core Babel team uses Linux, but we've had reports of the Babel server running in a Windows environment. Here are some notes:
- Tell MySQL to user UTF-8: patch for dbconnection.class.php
- Don't use the PHP crypt() function for authentication: bug 242011
Submitting code patches
Server code patches must be attached to Bugzilla bugs and enhancement requests. Please ensure patches are submitted against a recently updated HEAD stream. Look for org.eclipse.babel/server in:
Committer Development Process
Babel committers develop new code, and work on bug fixes on their local development environment. Only when a code set is complete can it be committed to the server for staging.
Before doing any work on the server code, please follow these simple guidelines:
- Make sure there is a bug that defines the scope of your work
- Implement the change on your local development environment
- Submit a patch for peer review
- Commit your code. Use the following syntax for the first line of your commit comment:
 Description of the bug
- Test the code on the staging area. CVS commits are published to the staging area within a minute.
- Request a release (see below) when it's appropriate.
Submitting database changes
Database versioning is something we are working on for the next Milestone release (Milestone 1). Currently if you want to make changes to the live database you need to put the SQL queries for alter/insert/update/create into the schema_change.sql file. When the next tagged version of the code is released the web master will process the schema_change.sql file by hand and make the appropriate changes to the database and test files.
Babel Server code in CVS is published to http://babel.eclipse.org/staging automatically every minute. To deploy the code to the live site:
- Tag the CVS code with a release tag: R_x_YYYYMMDDHHMM, where x is the release number and YYYYMMDDHHMM is the local time of the tag.
- Send an email to email@example.com to announce the request for a deployment, specifying the release tag.
The Designated Deployment Drone (currently Denis) then establishes an SSH connection to the Babel server, and as the deployment user 'genie', runs the ./deployBabelLive.sh <rtag> to deploy the release.
The Babel server has two data bridges:
- Eclipse Bugzilla: this bridge is used to maintain a mapping of the Bugzilla user accounts on the Babel server. A process on the Bugzilla server and on the Babel server monitors this URL for changes, triggering an export (and an import) of the data.
- Eclipse Project/Version info: When the Bugzilla bridge is run, the Project/Version info is pulled from this URL.