In v0.9 we would like to add a support for multi language cms page contents

• ticket:131
•  Forum thread

Here we are collecting some facts about this here.

First of all

• All i18n stuff should be easy to use
• it should not be too much work it to implement
• too much to configure works against "easy to use"
• we should try to give the user a straight workflow
• Easy to update for a existing installation

the facts

• default language:
  • the user must set a default language. How? (see below)
  • A page must at least exist in the default language
  • A new page is provided first in the default language, after this a user can create the page content in a other language
• split "Page" table into text content and meta data: All continuous data remain in the PAGE table. All language specific data went into a separated model.
• The shortcuts are always in the default language.
• menu/sitemap:
  • items change the language
  • pages which don't exists in requested language, should show up in default language

open facts

• How can the user specify the "other language" of a new page? Should we get him a list of all possible language? The  django global_settings.py holds a list of language.
• How can the user specify the default language? Should we use settings.LANGUAGE_CODE for this?
• language separation for templates too? (make sense if the user has some text in the template)
• Use the language specific PageContent class for the page archive, too?
• How to update? Ideas:
  • Only rename the old model class and add the new models. So we can access both models the old and new and migrate the data.
• shortcuts/permalinks:
  • Build shortcuts/permalinks only in the default language: The URL line and every permalink is always the same
  • use the requested/existing language: URL line/permalinks exists in different language.
• How to cache?
  • We have two i18n things:
    • i18n page content
    • i18n gettext strings in templates/plugin output (django has over 40 translations)
  • solutions:
    • Bad idea, blow up the cache with all variants: cache_key = url + cms_page_lang + client_lang
    • Other idea: Use the cms page language for gettext translation: connect cms_page_lang with client_lang

How to build urls/permalink?

current implementation:

• urls:
  • The current urls build from the page shortcut: domain.tld/shortcut1/shortcut2/
  • There exist no filename in a url
  • Only the last existing shortcut used to get the current page. All other parts of the url would be ignored (See also forum thread)
• permalink:
  • Permalinks looks like this: http://www.pylucid.org/_goto/125/internal-links/
  • Only the id in the url use to get the current page. The _goto view makes a redirect to the right page url.

The new implementation must recognize the existing language of a page.

• variant 1 - file based on page shortcut
  • urls:
    • There exist only one language independently shortcut for a page.
  • permalink:
    • Only the id in the url use to get the current page.
    • From the "filename" only the lang code are recognized all other parts ignored
  • pros
    • relative short urls
    • default language with, e.g:
      • domain.tld/shortcut1.html or domain.tld/shortcut1/ ?
  • cons
    • complicated structure
  • urls e.g.:
    • domain.tld/shortcut1.de.html
    • domain.tld/shortcut1.en.html
    • domain.tld/shortcut1/shortcut2.de.html
    • domain.tld/shortcut1/shortcut2.en.html
  • permalink e.g.:
    • domain.tld/_goto/125/shortcut1.de.html
    • domain.tld/_goto/125/shortcut2.en.html

• variant 2 - file based with a 'index' file
  • pros
    • default language with:
      • domain.tld/shortcut1/index.html or domain.tld/shortcut1/
  • cons
    • complicated structure
    • long
    • 404 on wrong url: domain.tld/shortcut1/ ?
  • urls e.g.:
    • domain.tld/shortcut1/index.de.html
    • domain.tld/shortcut1/index.en.html
    • domain.tld/shortcut1/shortcut2/index.de.html
    • domain.tld/shortcut1/shortcut2/index.en.html
  • permalink e.g.:
    • domain.tld/_goto/125/shortcut1/index.de.html
    • domain.tld/_goto/125/shortcut2/index.en.html

• variant 3 - language as the first directory
  • pros:
    • short
  • cons:
    • how to get the default lang?
      • domain.tld/shortcut1/ ? (difficult to keep apart)
  • urls e.g.:
    • domain.tld/de/shortcut1/
    • domain.tld/en/shortcut1/
    • domain.tld/de/shortcut1/shortcut2/
    • domain.tld/en/shortcut1/shortcut2/
  • permalink e.g.:
    • domain.tld/_goto/125/de/shortcut1/
    • domain.tld/_goto/125/en/shortcut2/

variant 4b is favoured

• variant 4 - language as the last directory
  • pros:
    • short
    • easy to get the default language (or we can built a page with a list of all available language):
      • domain.tld/shortcut1/
      • domain.tld/_goto/125/shortcut1/
  • cons
    • ?
  • urls e.g.:
    • domain.tld/shortcut1/de/
    • domain.tld/shortcut1/en/
    • domain.tld/shortcut1/shortcut2/de/
    • domain.tld/shortcut1/shortcut2/en/
    • variant 4b - lang.code as a prefix:
      • domain.tld/de/shortcut1/
      • domain.tld/en/shortcut1/
      • domain.tld/de/shortcut1/shortcut2/
      • domain.tld/en/shortcut1/shortcut2/
  • permalink e.g.:
    • domain.tld/_goto/125/shortcut1/de/
    • domain.tld/_goto/125/shortcut2/en/

• variant 5 - unique shortcuts
  • The page shortcuts are language dependent.
  • pros
    • short
    • permalink system doesn't change
  • cons
    • problem with not unique page names (e.g. anglicist pages like "links")
      • solution 1: Append language code on double shourtcuts, e.g.:
        • domain.tld/links-de/
        • domain.tld/links-en/
      • solution 2: urls always start with the language code as a "prefix", e.g.:
        • domain.tld/de/links/
        • domain.tld/en/links/
    • We can't make a "index" page, with a list of all available language.
  • urls e.g.:
    • domain.tld/contact/ - (en)
    • domain.tld/kontakt/ - (de)
  • permalink e.g.:
    • domain.tld/_goto/123/contact/
    • domain.tld/_goto/456/kontakt/

future plans

• If Django supports  Routing according also to hostname it's possible to use sub domains with language code, e.g:
  • /en/contact/ -> en.domain.tld/contact/
  • /de/contact/ -> de.domain.tld/contact/

links

• the new model file in the SVN trunk:  new_v09_models.py
•  http://pylucid.org/_goto/125/internal-links/
• django-multilingual:
  •  http://code.google.com/p/django-multilingual/
• transdb:
  •  http://code.google.com/p/transdb/
  •  http://transdb.googlecode.com/svn/trunk/
• django
  •  http://www.djangoproject.com/documentation/i18n/#how-django-discovers-language-preference
  •  http://www.djangoproject.com/documentation/settings/#language-code
  •  http://code.djangoproject.com/browser/django/trunk/django/conf/global_settings.py#L35
• i18n in urls:
  •  http://googlewebmastercentral.blogspot.com/2008/08/how-to-start-multilingual-site.html
  •  http://www.google.com/support/webmasters/bin/answer.py?answer=66359

---

(Important: If you want to edit wiki pages, please login as guest/guest.)