Multi-context & multi-page environment

[mrdimidium]

First of all, thank's for the amazing project!

I would be interested about multiple pages and page persistence support (cookies, localStorage, etc.).

The simplest example that comes to mind: we want to test that collaborative editing works in Google Docs. We open two tabs (they share a single authorization cookie), change the text in the first, and check that the changes have been applied in the second.

In the current version, we could spin up two instances, but then we can't verify that local synchronization is working (like Logux, which uses a localStorage subscription).

I guess your cloud runs multiple instances, and the load balancer allows to connect to a free instance. But it seems that this approach requires building your own balancer, which in the case of Chromium is the browser process.

It also seems that without support for multiple JS environments, you can't support service workers, web workers, and Chrome-style extensions.

I understand that you work with a small team with limited resources, I'm just interested to know your long-term plans.

P.S. Related to #882

[krichprollsch]

Hello @mrdimidium,

Multi-context and Multi-page is in one of our goal for the near future (~6 months), but it requires important changes in the browser.
The main one is the way we are using libdom, for now we can't have two DOM tree open at the same time.

[mrdimidium]

Sounds great! Am I correct in understanding the problem that libdom has some global objects that prevent multiple instances from running in a single process?

Do you think the lightpanda implementation will be multi-process (like blink/gecko/servo/webkit2) or will all pages run in a shared process? Will pages have their own sandboxes, or is isolation implied at a higher level (for example, in a container)?