| 开发者 | brianmaloney |
|---|---|
| 更新时间 | 2026年11月21日 10:20 |
| PHP版本: | 7.4 及以上 |
| WordPress版本: | 7.1 |
| 版权: | GPLv2 or later |
| 版权网址: | 版权信息 |
_redirects file and a _headers file (so feeds keep their type on Netlify and Cloudflare) are generated for you.
For developers
Every stage has actions and filters (list of hooks), and deploy targets are pluggable through a small interface. All features are available from WP-CLI: wp glaciate build, status, why, deploy, zip, settings, target, secret, schedule, forms, doctor and more.
define( 'GLACIATE_CONNECT_TO', '127.0.0.1:80' ); in wp-config.php to an address that reaches the web server.No. Every feature is free and stays free.
No. It makes requests to your own site, to render pages, and to the deploy target you configure (for example your S3 bucket or Netlify site), to publish. Nothing else, and nothing until you set a target up.
Encrypted in the database, with a key derived from the security keys in wp-config.php, and never displayed again. You can also set any of them with a constant or an environment variable (for example GLACIATE_S3_SECRET_ACCESS_KEY) so they are never stored at all. If the security keys change, saved credentials can no longer be read; Site Health says which ones to enter again.
No. On a folder, SFTP server or S3 bucket it only deletes files it published there itself, and a folder must be empty or marked as its own first. A Git branch or Cloudflare Worker that already has content from elsewhere is refused unless you allow replacing it.
Yes. Glaciate publishes a search index with your site, and your theme's search forms open a results window that searches it in the browser. Results include every published page (title and text from the page itself), and the index is updated along with the pages that change. You can turn it off under Settings → Search.
Yes, if your host builds the site: a GitHub Actions workflow on the Git target, Netlify or Cloudflare Pages. Add npx pagefind --site . to that build and choose "Pagefind, run by your host's build step" under Settings → Search. Glaciate doesn't include or run Pagefind, because it is a separate program. Your search forms keep the same results window. Themes can choose what is searchable by marking elements with data-glaciate-search, or leave parts out with data-glaciate-search-ignore.
Yes, once you turn on Forms under Settings → Forms. A static host can't run them, so Glaciate points them at a receiver:
wp glaciate forms receiver php|cloudflare|netlify): a single PHP file for any PHP host, a Cloudflare Worker or a Netlify Function. It checks each submission and forwards it to your site, signed with a secret you set on both sides. It can also send submissions to a webhook, or (the PHP file) by email, without WordPress.data-glaciate-form="name"; their submissions are emailed to you. File uploads aren't supported.
Every form gets a hidden field that people don't see and bots fill in, the time the page was open is checked (bots are quicker than people), each visitor can send five forms in ten minutes, and submissions must come from your static site. For more protection choose Cloudflare Turnstile or hCaptcha; these are outside services, so they are off unless you pick one.
Drafts, private posts and password-protected posts are never published. You can leave out more addresses in Settings.
$_SERVER, and it empties $_POST and $_REQUEST afterwards.GLACIATE_…), the REST namespace (glaciate/v1) and the WP-CLI command (wp glaciate) are renamed to match.wp-content/uploads/glaciate/ instead of a folder directly in wp-content.wp glaciate zip and wp glaciate forms receiver no longer take --out. wp glaciate zip --porcelain prints just the archive's path, and forms receiver prints the receiver, so redirect it to a file.wp glaciate changes preview and the REST route glaciate/v1/changes.wp glaciate doctor._redirects only.data-glaciate-form post to a receiver, and WordPress hands each submission to the form plugin. Answers and field errors appear in the form; without JavaScript a short page answers.?s= links open a results window; only the index files a query needs are downloaded, and only the parts of the index that changed are redeployed.