https://bugzilla.mozilla.org/show_bug.cgi?id=1288684

https://bugzilla.mozilla.org/show_bug.cgi?id=1306011

bug 1288406

• Inline code comments
• docstrings
  • src/lib/dom
  • src/lib/observer
  • src/lib/format
  • src/intl
  • src/ftl
• usage docs (restructured text, gecko RTD)

https://bugzilla.mozilla.org/show_bug.cgi?id=1288690

https://bugzilla.mozilla.org/show_bug.cgi?id=1288443

https://bugzilla.mozilla.org/show_bug.cgi?id=1303883

https://bugzilla.mozilla.org/show_bug.cgi?id=1291915

https://bugzilla.mozilla.org/show_bug.cgi?id=1280669

https://bugzilla.mozilla.org/show_bug.cgi?id=1280685

https://bugzilla.mozilla.org/show_bug.cgi?id=1280891

https://bugzilla.mozilla.org/show_bug.cgi?id=1280901

For localizers using pontoon, we'll need to document selectors and traits.

This will depend on how much the pontoon UI is helping with each of the facets of l20n at the time strings hit aurora/pontoon, see issue #14.

elmo uses django, pontoon jinja

• FTL support, bug 1280891
• value checks bug 1280893

There are still a couple items left from the current ECMA402 spec that we don't support.

https://bugzilla.mozilla.org/show_bug.cgi?id=1280893

Input Data

The Migration process uses

• en-US ftl
• l10n ftl existing state
• legacy l10n data in DTDs, properties, etc
• Operations

Operations

The current list of operations are

• COPY with unescaping of < and \u00a0
• PLACABLE REFORMAT, &brandShortName;, %S, %1$S, { name }, #1
• PLURAL reconstruction, analyze existing plural string, map PluralForms.jsm numbers to CLDR keys, iterate over strings, create selector
• CONCAT all of the above, plus literals
• DELETE id_or_trait, in case we want to delete existing messages or one of their traits

Message Generation

Messages are generated with an extended AST, which creates the AST of a message with the operations and data above.

Message Merge

Message merge takes the new (and to be deleted) messages, along with their attribution. It then executes a merge algorithm for each author and commits that.

The merge algorithm takes the en-US ftl as template, to create sections, comments, and order messages. It creates a minimal edit based on top of the (existing or newly created) l10n ftl file. It doesn't reformat outside of messages that are given by the operations for the current author.

In case of doubt, strings are inserted last into the section they belong to.

Much of this will be implemented as part of python-l20n.

• parser
• resolver
• Localization
• DOM
• Mochi tests for mozilla-central

Elmo uses django-nose, nose, and a testsuiterunner based on django_nose.NoseTestSuiteRunner.

The base testcase is django.test.TestCase.

@mathjazz, what does pontoon use?

When we have a good understanding of our potential performance impact, we need to come to a decision with the rest of the org on next steps.

https://bugzilla.mozilla.org/show_bug.cgi?id=1288146

https://bugzilla.mozilla.org/show_bug.cgi?id=1311666

https://bugzilla.mozilla.org/show_bug.cgi?id=1280683

https://bugzilla.mozilla.org/show_bug.cgi?id=1280690

We'll need to document how to localize FTL. Target audience are localizers on central using text editors.