./blog.sh 1.12

./blog.sh v1.12 — part 1 of 2

./blog.sh 1.12 is out — release on GitHub.

The translation release. ./blog.sh translate <slug> --lang <code> --auto hands a post to a translator you name in env.sh: a program of your own, OpenAI and everything that speaks its API, Anthropic, Gemini, DeepL or LibreTranslate. It leaves a draft for you to read and save, and saves nothing by itself. Before writing the file it compares the translation with the original in addresses, pictures, numbers and the count of headings and paragraphs, and says what differs. ./setup.sh --translator sets it up, and the Android and desktop apps have a button for it since 0.10.0. A first draft in the other language →

Beside that, a reader can reply to a toot from their own Mastodon server: one field under the comments, remembered in their browser, nothing of it sent to your site. Reply from your own server → And a post is announced once its page is live, not three minutes before. A deploy that fails leaves the toot owed for the next one, and doctor names it. A hashtag keeps a tag's words together, so blog.sh is #blog_sh, not #blog. The toot comes after the page →

Most of the release started as a report from Pavel: the translation itself (#68), the reply from another server (#75), the announcement that came too early (#83) and four smaller ones (#79 to #82). Thank you. The rest is in everything else in 1.12, doctor warning about a Ruby that no longer gets security fixes among it.

Nothing to migrate — git pull, rebuild, deploy. Hashtags of tags with a dot, a hyphen or a space change shape, and a site with no deploy target is no longer announced. Full notes →

Comments