Built-in transitions disable unmount

[yomotsu]

Describe the bug

A child component with a built-in transition, using a store value for the transition condition will not be unmounted.
(This problem is even worse in svelt-kit. the "+page" instance will not be unmounted and duplicated endlessly)

Workaround:
Use custom transitions instead of built-in transitions.

Screen.Recording.2022-11-07.at.12.35.03.AM.mp4

Reproduction

1. Open REPL: https://svelte.dev/repl/c5b7ff813f7f452683cf49740a7f8323?version=3.52.0
2. Press the page1 button. Confirm "page1 contents" will be mounted.
3. Press the page2 button. Confirm "page1 contents" will be unmounted, and "page2 contents" will be mounted.
4. Press the page1 button again. Confirm "page2 contents" will not be unmounted, and "page1 contents" will be mounted.

Logs

No response

System Info

System:
    OS: macOS 12.6
    CPU: (8) x64 Intel(R) Core(TM) i7-1068NG7 CPU @ 2.30GHz
    Memory: 5.52 GB / 32.00 GB
    Shell: 5.8.1 - /bin/zsh
  Binaries:
    Node: 16.17.1 - ~/.anyenv/envs/nodenv/versions/16.17.1/bin/node
    Yarn: 1.22.18 - /usr/local/bin/yarn
    npm: 8.15.0 - ~/.anyenv/envs/nodenv/versions/16.17.1/bin/npm
  Browsers:
    Chrome: 107.0.5304.87
    Chrome Canary: 86.0.4196.0
    Firefox: 106.0.1
    Safari: 16.0
    Safari Technology Preview: 16.4
  npmPackages:
    svelte: ^3.52.0 => 3.52.0

Severity

annoyance

[eltigerchino]

A workaround is to add local to the transition such as <div transition:fade|local>

I've personally come across this problem many times and seen others report the same problem. Perhaps some kind of change would be good to prevent this.

[CodingWill0]

Seems like Svelte has always had issues with transitions. I confirm I had this double onMount issue after a native Svelte goto method and was fixed adding local directive. Given that local directive has been a fix to multiple Svelte issues in the past related to transitions, maybe it should become the default behaviour of transitions. #6812

[yomotsu]

@s3812497 @CodingWill0
Thanks for the workaround, and the link for the existing issue.

Seems this report is duplicated of #6812.
Let me close this in favor of that.