A public channel for the Angular community to come talk about Scully and JAMStack strategies in Angular.
Hi all,
we have a highly dynamic web site that loads content from GraphCMS. Therefore we can't do this once on the CI. To always have an up-to-date static site, a cronjob runs the 5 minutes and run ng build and npx scully -- --no-warning. We use the hostUrl setting to actually scrape the live site, instead of using the offline generated build output files from the angular build.
To make this work, we had to set outputHashing=none, otherwise our regular angular app (running in a docker container) and the static site would have different *<hash>.js files (and also other resources) referenced in the generated HTML.
This would lead to an evergrowing amount of files on the CDN too. Futhermore it leads to other problems when scully tries to boot the regular angular app.
Unfortunately there are not infos related to outputHashing, to using the hostUrl setting, or how to actually do what we do.
Our way of doing things "kind of" works - the only issue we are facing is caching related. As the JS resources (runtime.js, ...) are unversioned, the browser doesn't always load the new resources.
The main problem seems to be that the scully build process generates new angular build files and copies those files to the static output folder, instead of using the resources from the site specified via hostUrl. I know I could write a plugin that downloads all resources, but it feels that should be a functionality of scully itself. Or why is it even possible to scrape a site - but only "half-way".
Hope you can help me.
cheers matti
The local files are used by design. We don't want to enable use-cases where Scully is used to copy-cat a site.
18 replies
registerPlugin('postProcessByDom', 'addMyCss', async (dom, route) => {
const { window: { document } } = dom;
const style = document.createElement('style')
style.innerText = `/* css content here */`
document.head.appendChild(style)
return dom
})defaultPostRenderers
