07 Jul 2025
Jekyll-Polyglot 1.10 is now available. It has vast improvements and changes to the i18n_headers
liquid plugin for SEO improvements, and minor adjustments for parallel build idempotency. Community Contributions and Vibe Coding helped with a large portion of these release features, testing and blog post.
This version of polyglot now requires ruby >= 3.1 , which may break in build systems.
the i18n_headers
plugin now has extended capabilities as part of this release:
- it will add
<link rel="canonical" ...>
for each page language, so indexing is unique across sites.
- it will add
<link rel="alternate" hreflang="x-default" ...>
to point to default language version of a site, when no matching language is requested by the browser.
- it will correctly define
<link rel="alternate" hreflang="..." >
for pages and posts in collections with custom permalinks.
- the default url will now include the
site.baseUrl
if defined
this additionally fixes a bug noticed that caused absolute url relativization to mangle these tags unintentionally.
vibe-coded contributions
Using certain vibe coding tools have helped find, measure and verify bug fixes and features for this release.
This is a new approach for software development, and it created advanced ruby tests against jekyll plugin code running against many built site languages.
The tests written with vibe coding helped ensure the code coverage remained high, and complex features could be added with confidence. Ensuring the test automation was in place ensured tricky features could be built correctly.
Additionally vibe coding tools helped translate this blog post into many languages.
Jekyll-Polyglot has been supported by humans. Human language documentation is contributed by humans who want to see this plugin documented in their native tongue. Humans contributing bug fixes and documentation are what have helped this plugin reach thousands of downloads a release. AI assisted programming, in my hands or yours, will shape the software we use, and in the many languages we write and speak.
ruby >= 3.1 required
Ongoing security updates to jekyll-polyglot build-time dependencies required a major version upgrade to ruby 3.1 . This may affect build systems that build their docs with jekyll-polyglot. But now is a good time to upgrade to the latest ruby major by now. Speak up if these changes cause complications with jekyll builds.
18 Jan 2025
Jekyll-Polyglot 1.9.0 has been released, which has minor dependency updates, and instructional improvements for getting the most from your multi-language website.
Thank you to aturret for helping to maintain the existing zh-CN
site pages. 谢谢!
george-gca improved the optional derive_lang_from_path
configuration to better identify document language from the path inference. Tests were added for his helpful feature improvement PR. This improvement helps infer the language of posts and pages missing lang
frontmatter, from any part of the document filepath.
Github user yunseo-kim submitted a instructions to improve sitemap generation . To help with SEO, a website should have only one root sitemap.xml , and not have duplicates for each sub-language site. Be sure to add the sitemap.xml
to the exclude_from_localization
configuration.
18 Aug 2024
Jekyll-Polyglot 1.8.1 has been released, which has a few feature improvements and recognizes community found bugs and provided fixes.
hacketiwack provided a stricter check for setting a doc permalink, preventing downstream problems with empty frontmatter fields.
Github user blackpill submitted a one character bugfix for the i18n headers tag when rendering the default language link alternative href.
17 Mar 2024
Get excited for Jekyll-Polyglot 1.8.0, which has a few feature improvements and recognizes community documentation and contributions!
language specific permalinks
One new feature is to give pages language specific permalinks and to retain their association to other relative pages. This new feature is again improved by antoniovazquezblanco, who is a gentleman and a scholar.
sitemap generation & i18n SEO
This release also recognizes the quality sitemap.xml and robots.txt solution provided by jerturowetz. This website now demonstrates and captures more SEO power by using these to be crawlable as a static jekyll website by search providers. See the example website files here.
jekyll :polyglot :post_write hook
Github user obfusk contributed a tiny PR a few years back:
With polyglot :site, :post_write
like these run for each child processes:
Jekyll::Hooks.register :site, :post_write do |site|
...
end
This release adds a custom :post_write
hook that runs exactly once, after all languages been processed (whether or not parallel_localization
is used):
Jekyll::Hooks.register :polyglot, :post_write do |site|
# do something amazing here!
end
This feature is helpful for complex jekyll static sites that make additional use of jekyll hook plugins.
She also contributed a fix for additional logging when language subprocesses crash. Thanks for this contribution!
localized variables and portuguese translation.
george-gca is a talented and awesome guy, contributing an entire blogpost on how best to localize rich text from site data. He also provided a site brazilian translation.
29 Feb 2024
Polyglot allows you to have different pages for different languages in your Jekyll site. For example, one could have a page about.md
in English and another about.md
in Spanish with completely different layouts. But if you want to have the same layout for both pages, you can use localized variables. This is a way to have different data for different languages in your Jekyll site, but using the same layout for all languages.
As an example I will use a template site created with Polyglot.
Sharing a layout between pages
In that site they have an about page for every language, in their case english in _pages/en-us/about.md and brazilian portuguese in _pages/pt-br/about.md. In both pages we can see that they have the same keys in the frontmatter, but some with different values. Both files point to the same layout, about, and this layout uses the values in the frontmatter to render the page.
For example, the subtitle
key in the english page has the value subtitle: <a href='#'>Affiliations</a>. Address. Contacts. Moto. Etc.
and in the brazilian portuguese page it has subtitle: <a href='#'>Afiliações</a>. Endereço. Contatos. Lema. Etc.
. To use this information in the layout, it is used like this:
The same goes for the content below the frontmatter in both files, which is simply used in the layout like this:
Polyglot will automatically render the page with the correct values for the current language.
Sharing a layout between pages with localized data
For the subtitle
of the page they used key: value
pairs in the frontmatter, but sometimes we want to use these same pairs in different parts of the site. For example, if we want to use the same subtitle
in the about.md
and in another page, we would have to repeat the same pair in the frontmatter of both pages. This is not ideal because if we want to change the subtitle
we would have to change it in two places. This is where localized data comes in. You can create a file like _data/:lang/strings.yml
, one for each language, and Polyglot will bring those keys under site.data[:lang].strings
.
For example, in the template site there are two files, _data/en-us/strings.yml and _data/pt-br/strings.yml. In the first file they have:
latest_posts: latest posts
And in the second file they have:
latest_posts: últimas postagens
This way, they can use the latest_posts
key in the layout like this:
{{ site.data[site.active_lang].strings.latest_posts }}
Which will correctly get the value for the latest_posts
variable defined in the file _data/:lang/strings.yml
for the current language.
Defining which variable to use in the frontmatter
Now if you want to define this variable in the frontmatter of the page, this gets a little bit trickier. One possible solution is to check if the value of the variable has a .
in it, and if it does use the value in the file _data/:lang/strings.yml
. This is how you would do it:
{% if frontmatter_var contains '.' %}
{% assign first_part = frontmatter_var | split: '.' | first %}
{% assign last_part = frontmatter_var | split: '.' | last %}
{% capture result %}{{ site.data[site.active_lang].strings[first_part][last_part] }}{% endcapture %}
{% endif %}
{{ result }}
This will work, for example, if frontmatter_var = blog.title
.
Now, if you need to check if the localization string (in this case blog.title
) actually exists in the file _data/:lang/strings.yml
before using it, you’ll have to create a plugin to check if the variable exists in the file _data/:lang/strings.yml
and if it does, use it, otherwise fallback to any value you want. I will not go into detail on how to do this, but I will show you how to use it. You can see the code for the plugin here.
{% if frontmatter_var contains '.' %}
{% capture contains_localization %}{% localization_exists {{ frontmatter_var }} %}{% endcapture %}
{% if contains_localization == 'true' %}
{% assign first_part = frontmatter_var | split: '.' | first %}
{% assign last_part = frontmatter_var | split: '.' | last %}
{% capture result %}{{ site.data[site.active_lang].strings[first_part][last_part] }}{% endcapture %}
{% else %}
{% capture result %}fallback value{% endcapture %}
{% endif %}
{% endif %}
{{ result }}