The stencil duplicator or mimeograph machine (often abbreviated to mimeo) is a low-cost duplicating machine that works by forcing ink through a stencil onto paper.wikipedia
As a front-end developer it’s a pain to write code in one way when you add CSS/JS in a back-end CMS and in a different way when you add them to your front-end app, gatsby solves this by basically creating a hydrated version of a ReactJS app with data provided from a build-time GraphQL server (I’m over simplifying here, but hey). This makes front-end developers happy because templates look and feel like React components (sort of) and back-end developers are happy because there’s a declarative way of exposing the whole taxonomy and data of a site through a common API (GraphQL).
As awesome as GatsbyJS and ReactJS are I want to re-invent the wheel because:
One thing I have to admit that GatsbyJS got right is that you want a unified data-access pipeline and that GraphQL is the correct protocol for it, but I’m not 100% convinced that putting all your logic there is the correct thing. So the idea is to use GraphQL to fetch content for pages, and for this GraphQL server to only have to be avaliable during build time.
It’s trivial to set upp a StencilJS app that uses
stencil-router or similar to parse parameters from the URL and fetch some data over GraphQL to display on the page, actually this is how most of us write a basic front-end application today. But what pages to I need to render?
In GatbyJS you get this fancy
gatsby-node.jsfile that you can use to call the internal API’s to tell the build server what pages should be generated, however flexible this is I just find it annoying.
There’s an old-school way of defining possible routes on your web site, a
sitemap.xml. So the default way is just to make one of those and the rendering process will use that as input. If that’s not enoug I’m thinkning a
sitemap.ts would be easy enough to implement so that whoever want more flexibility can get it.
Possibly. I’m thinking that injecting a different GraphQL client during SSR than post-hydration to get a division between static/dynamic content without having to do anything with your client code.
Another thing would be serializing any access data during SSR into the HTML but in such a way that it would populate something like
apollo-cache-inmemory so data access done after re-hydration is available without reloading.
Stole it from here Where I found out that 謄 (if I understand it correctly) means “Mimeograph” - perfect.