New extension: Volumes (previously CobaltVDisk)

[kx1bx1]

Volumes development is almost finished!

Volumes' (previously CobaltVDisk) development has been completed, and I am now moving on to testing all of its features. Right now, it's only available on my extensions gallery (but since I don't want to self-advertise, I won't link it here) as a beta. However, once it's finished, I'll commit the completed code here and mark it as "released" in my gallery.

Note: For some reason, Canva's SVGs can't render properly, so I had to export them as PNGs.

Volumes is an enhancement of rxFS by 0832. Its main focus is speed improvements and new features like file size limits, permissions, and more.

The versioning scheme

Volumes uses a structured, time-based versioning scheme; full details will be documented once implementation is complete. TL;DR: You don't have to read this.

Volumes go by a '3-monthly' update schedule. If any changes to Volumes were made by me, a PR will be made as a 'cumulative update' to Volumes. As of right now, Volumes' versioning schema has been removed, as its target audience probably does not care what version Volumes is on.

Why use the MIT license?

My choice of using the MIT license instead of the GPL-3.0 license is not only due to brevity, but also due to the fact that I want Volumes to be aligned with TurboWarp’s existing ecosystem and extension practices. MIT is rather open and easy to understand for users, and it's quite permissive, so that's what I used.

Full disclosure: this extension was LLM-generated

TL;DR: You don't have to read this if you do not care about LLM-generated code. If it writes bad code, it won't be accepted anyway.

This extension's code is purely LLM-generated. When I say this, I do not mean I only generated it in one prompt. I bring my models to a high standard, and you could probably count 50 prompts that assess security, stability, and speed with Volumes. If AI-generated code is used, it should meet or exceed human-quality standards.

Models used:

1. Primarily Gemini 3.0 Pro
2. Gemini 2.5 Pro (initial development)
3. GPT 5.0, 5.1, and 5.2 (bug testing)

I do intend to do the 'ultimate test' of Volumes by creating a project that primarily uses it, to ensure I can mark this PR ready to review.

Why so few commits?

This PR is planned to only have 5-10 commits because one commit is supposed to be one Volume update after development is finished. This PR will be squashed or rebased once development is complete.

About this PR

• This extension shows properly in the Extensions Gallery
• This extension was properly linted
• This extension was properly formatted
• This extension was properly validated
• Development for this extension is finished

These are relatively easy checks to pass because, well, the 'extension' code is literally just comments until I finish development.

Planned features

In this context, "Completed" does not mean "Added to this branch". It means that the feature is done on my side. I don't commit new features to this PR.

Feature	Priority	Status
Testing	High	Paused
Possible new features	High	Completed
Core filesystem operations	High	Completed
Filesystem import and export	High	Completed
Permissions system	High	Completed
New features	Medium	Completed
User experience improvements	Medium	Completed
Extension banner	Medium	Completed
Extension documentation	Low	Completed
Per-sprite permissions (ACL)	Medium	Scrapped

[8to16]

@kx1bx1 please stop generating and submitting AI-generated extensions to the gallery

(note: this is my personal opinion; see #2260 or maybe see if @GarboMuffin has anything to say)

Edit: just thought about this comment again and it feels plain rude; but your extension does not contain any actual code; just a few comments. I have read the comment explaining this in your code, but IMO you should only make a PR when your extension somewhat works (the mesh PR I recently made in #2382 does somewhat work because messages are indeed being sent over the broadcastchannel, but the hat block doesn't work due to fields vs inputs.)

[kx1bx1]

@kx1bx1 please stop generating and submitting AI-generated extensions to the gallery

(note: this is my personal opinion; see #2260 or maybe see if @GarboMuffin has anything to say)

Edit: just thought about this comment again and it feels plain rude; but your extension does not contain any actual code; just a few comments. I have read the comment explaining this in your code, but IMO you should only make a PR when your extension somewhat works (the mesh PR I recently made in #2382 does somewhat work because messages are indeed being sent over the broadcastchannel, but the hat block doesn't work due to fields vs inputs.)

Yes, this PR does not have any actual code. The reason I haven’t posted the source yet is personal, and a PR isn’t a good venue for that, so I’ll omit it here.

I agree that PRs should be reviewable and functional, and I’ll share Volumes’ source once it’s ready to demonstrate working behavior. If LLM-generated code is ever used, and I've said it before, it should meet or exceed human standards.

[GarboMuffin]

If you have no code that you want reviewed then you shouldn't make the pull request yet

[GarboMuffin]

GitHub should let you make a new pull request for when you have something to get looked at

[8to16]

My choice of using the MIT license instead of the GPL-3.0 license is not only due to brevity, but also due to the fact that I want Volumes to be aligned with TurboWarp’s existing ecosystem and extension practices. MIT is rather open and easy to understand for users, and it's quite permissive, so that's what I used.

as per CONTRIBUTING.md you can use MPL-2.0 instead, which is a good balance, though this is not enforced

[kx1bx1]

My choice of using the MIT license instead of the GPL-3.0 license is not only due to brevity, but also due to the fact that I want Volumes to be aligned with TurboWarp’s existing ecosystem and extension practices. MIT is rather open and easy to understand for users, and it's quite permissive, so that's what I used.

as per CONTRIBUTING.md you can use MPL-2.0 instead, which is a good balance, though this is not enforced

I might do either GPL-3.0 or MPL-2.0; if I can't choose, I might as well do a coin flip.