doc: clarify Web Storage behavior

[cjihrig]

Affected URL(s)

https://nodejs.org/api/globals.html, https://nodejs.org/api/cli.html

Description of the problem

I think that we should clarify some behaviors of Web Storage in Node.js vs. browsers.

• Session storage and local storage are both scoped to the current process (not individual users or server requests). This is important for applications like server side rendering.
• Local storage uses the value of --localstorage-file as its origin.

Other things that would be good to note:

• Local storage files can be accessed by multiple processes simultaneously.
• The local storage quota is also 10MB. This is noted for session storage.

For reference, here are the Deno docs on Web Storage.

[ameer-w-ali]

Hello sir, can you assign this Issue to me. I want to start my contribution journey and I think I can handle this issue

[cjihrig]

We don't typically assign issues. If you want to work on this, just open a PR 😄

[Rekl0w]

PR submitted.

[jakecastelli]

Hi Colin, a quick question here, I am trying to figure out where the 10MB quota comes from in node (if you don't mind pointing out for me).

Deno has its own implementation of webstorage api and the 10mb comes from const MAX_STORAGE_BYTES: usize = 10 * 1024 * 1024; in denoland/deno/ext/webstorage/lib.rs here, in nodejs the closest thing I could find its SQLITE_CONSTRAINT is 19 which means /* Abort due to constraint violation */, maybe I missed something.

[cjihrig]

Sure. It comes from here.

[jakecastelli]

Ohh I see, thanks! A follow up question, should there be a test case for it?

[cjihrig]

There are WPT tests for both local and session storage. They don't test for a specific quota value though, as that part is left up to individual implementors.

[jakecastelli]

Thanks for giving more context on the WPT tests!

I think I should've asked a bit more specific, should there be a test for testing 10mb quota in node? Given the fact it is up to individual implementors. Hope my question is not disturbing though 🙏

[cjihrig]

I guess it wouldn't hurt. I would recommend not doing repeated writes like the WPT tests though. That approach works because they don't know what the upper bound will be. Since we know Node's upper limit, we can do it in just a few writes.