everzet's blog

Behat 2.3-dev + Gherkin 1.2-dev

I’ve just finished i18n refactoring for both Behat and Gherkin. Now they both use simple i18n.php files containing simple translation arrays. This switch gave both Behat and it’s Gherkin parser noticeable performance boost during startup.

Behat’s i18n.php

Behat always had its own translation dictionary, used by its formatters. Every message, that Behat prints to you, like failed steps or 12 scenarios failed could be translated with --lang option and translations for this lived inside i18n/ directory. It was *.xliff files. This means, that every time you start Behat, it parsed those files with DomXml, which took some time. Even more than that, xliff is not natural for lot of people and i think it’s the main reason of low level of contributions here.

Now every single Behat translation live inside one simple PHP file. It’s easier than ever now to contribute - just send a PR with this simple file change and in next release Behat will talk with you in your language.

I’ve transited every translation from old scheme to new one, so everything here should work as expected. Yeah, should…

Gherkin’s i18n.php

Gherkin have more important change. As with Behat, now every translation lives under the i18n.php file, but in opposition to Behat’s i18n.php Gherkin’s catalogue is loaded/autogenerated from the cucumber/gherkin translation dictionary. It means, that from now on, every *.feature file written for Behat in any of supported 40 languages would be valid across all languages and Gherkin implementations. Want to migrate your project from PHP to Ruby? No need to rewrite *.feature files - you only need to redefine steps and Cucumber will parse your features succesfully.

Behat\Gherkin’s i18n.php now will be always in sync with cucumber/gherkin, means, that to fix/update some translation, you need to send Pull Request here.

It is a great switch for the Behat future, but it also have huge drawback. Behat\Gherkin’s i18n library have been diverged with Cucumber’s one. It means, that if you’ve used some different languages for your *.feature files, the chances are, that you need to fix keywords in them now OR send a PR to cucumber/gherkin

Test your suites

All changes are in develop branch of Behat. To test if your suite is works with it, simply install dev version of Behat:

$> git clone git://github.com/Behat/Behat.git
$> cd Behat
$> git checkout -b develop origin/develop
$> wget -nc http://getcomposer.org/composer.phar
$> php composer.phar install

After that, you will be able to run Behat with:

$> /path/to/Behat/bin/behat

If you like/don’t like something about those changes, comment in this post. This is the chance for you to change/fix something before Behat 2.3 and Gherkin 1.2 releases.