Fix: 2 islands on 1 page are revived

[sylc]

Fix: #196

The island end comment node was removed before the call to node.nextSibling, stopping the walk. I have re-ordered the deletion and added a test.

[lucacasonato]

Thanks for the fix. I think this should generally work, except in the case of nested islands. Nested islands will not work like you are expecting. In the test with <Inner> and <Outer>, the inner component is never directly hydrated. Instead it is hydrated as part of <Outer>. In this case this results in the correct behavior, but it may not always, as the props that are serialized into the HTML will not be used to revive the <Inner> island.

[lucacasonato]

I guess this does fix nested islands like I described in #168, so we can probably land this. The only concern is that there is a little bit of inefficiency because we are shipping the props for the <Inner> component, even though we are not hydrating it.

[sylc]

I guess this does fix nested islands like I described in #168, so we can probably land this. The only concern is that there is a little bit of inefficiency because we are shipping the props for the <Inner> component, even though we are not hydrating it.

This intend of this PR is not really to fix #168 but to fix #196 (the reviving issue when 2 island are in one page). I added to test for the nested island because i though they would be needed in the future anyway and to make sure there was no regression.

nested island still don't work for example with a thin outer as per reported in #178

if you think, that the nested island test should be added in a PR with a more holistic way around the nested island issue, then they can be removed.

[lucacasonato]

LGTM.

I have removed the nested islands test for now.

Thanks for working on this.