Option to disable client-side routing

[benmccann]

@ConProgramming wrote to support JungleJS as a SvelteKit adapter

Related: #231

[benmccann]

I'm not that familiar with the new testing code, but if there's a way to write a simple test for this that'd probably be nice

Also, this should probably go in the docs at least in https://github.com/sveltejs/kit/blob/master/documentation/docs/13-configuration.md

[benmccann]

I wonder if it might be better to make this routing: { client: false } or something in case we want to add other routing options. I'm curious what other folks think about names

[ConProgramming]

I thought about that maybe there should be an enum for routing, but Rich's original issue had it as clientSideRouting so I decided it'd be simpler to leave it as that

[GrygrFlzr]

Aside from the flexibility of adding more routing options, for intellisense purposes it might also be better to provide a routing object because then 'client?', 'server?', etc. options would all show up localized in the object.

[Rich-Harris]

I can envisage cases where you might want to disable hydration on specific pages that you know to be purely static. Are there cases where you might want to disable client-side routing on specific pages as well? i.e. does this make sense?

<script context="module">
  export const prerender = true;
  export const hydrate = false;
  export const routing = false;
</script>

The answer to that presumably influences whatever API we settle on.

Separately, a Svelte-ish thing to do here would be to treeshake the router away if routing is disabled for a given app (or indeed page).

[ConProgramming]

Being able to disable hydration on purely static pages would be nice - having the whole "0kb js" on static pages like elderjs can do.

My first thought is client side routing is all-or-nothing, but I guess someone could want authenticated pages with routing yet also static prerendered pages without it...

I thought about wanting to tree-shake away the router, but then I'd assume we'd want a simple reimplementation of "goto" to still work, just without csr?

[benmccann]

Closing in favor of #713