fix: duplicate @preact/signals version

[marvinhagemeister]

We initialized new projects with @preact/signals@2.x but Fresh itself was using 1.x. There shall only ever be one version.

Kinda makes me wish Deno had some sort of peerDependencies.

Fixes #2962

[csvn]

Kinda makes me wish Deno had some sort of peerDependencies.

Yes, and version ranges, e.g. v1.x || v2.x.

[csvn]

I think tests will need to wait microtask/animation frame to match what Signals v2 had as breaking change. 🤔

Edit: @preact/signals@2.0.0

preactjs/signals#604 fea3e8d Thanks @JoviDeCroock! - Defer all DOM updates by an animation frame, this should make it so
that any previously synchronous DOM update will be instead delayed by an
animation frame. This allows Preact to first perform its own render
cycle and then our direct DOM updates to occur. These will now
be performed in a batched way which is more performant as the browser
is prepared to handle these during the animation frame.

This does impact how Preact based signals are tested, when
you perform a signal update, you'll need to wrap it in act. In a way
this was always the case, as a signal update that resulted in
a Preact state update would require it to be wrapped in act, but
now this is the norm.

Edit 2: Perhaps using act() from preact/test-utils can help, if there are signal changes in tests?

To execute state updates and effects synchronously, use the act function from preact/test-utils to wrap the code that triggers the updates.

Edit 3: Nvm, seems it was another issue, multiple different signal versions that may have been the issue 😅

[marvinhagemeister]

Upstream Deno issue for peerDependencies outside of npm packages denoland/deno#26199

[marvinhagemeister]

I think tests will need to wait microtask/animation frame to match what Signals v2 had as breaking change. 🤔

Nah, only the www e2e tests were failing. Turns out I forgot to increase the version number for signals in that.

[csvn]

Is it not possible to remove this dependency from here instead? Will that not use the same version as the root deno.json instead via workspaces?

[marvinhagemeister]

Yeah there is a PR for that #2900 . Was kinda hoping we could have support for workspace:* syntax similar to pnpm in Deno by now, but it looks like that is getting pushed on the back burner.