SQLite encrypted additional package

[fileman]

Add SQLite encryption as additional package without 3rd party license required (see .net 11 breaking change)

[adrianhall]

This needs a thorough review, so it won't be a part of 10.1.0 - love the idea, though.

[adrianhall]

LLM Review:

The new encryption package is well-designed, well-documented, and genuinely proves its core claims through tests. The single headline issue is architectural, not cryptographic: swapping the core client to the bundle-less …Sqlite.Core provider is a silent breaking change that will compile fine but crash at runtime for every existing consumer that doesn't add its own SQLitePCLRaw bundle (P0).

Adrians Notes:

First of all, thank you for taking the time to present a well-thought out codebase. It's rare that the LLM reviewer does not have ANY security issues reported.

That being said, I am unwilling to break existing developers in this way. This one can't be accepted as is. I'm willing to discuss a refactor of the underlying client-side database for the v12 release (aligned with .NET 12) but not in a v10 release.

We should probably take up this design in a Discussion first so we can determine the best way forward for a v12 release. We can then bring some of this work (refactored) along for the ride then.

Until then, you should consider a fork of the library for encrypted SQLite instead.

[fileman]

LLM Review:

The new encryption package is well-designed, well-documented, and genuinely proves its core claims through tests. The single headline issue is architectural, not cryptographic: swapping the core client to the bundle-less …Sqlite.Core provider is a silent breaking change that will compile fine but crash at runtime for every existing consumer that doesn't add its own SQLitePCLRaw bundle (P0).

Adrians Notes:

First of all, thank you for taking the time to present a well-thought out codebase. It's rare that the LLM reviewer does not have ANY security issues reported.

That being said, I am unwilling to break existing developers in this way. This one can't be accepted as is. I'm willing to discuss a refactor of the underlying client-side database for the v12 release (aligned with .NET 12) but not in a v10 release.

We should probably take up this design in a Discussion first so we can determine the best way forward for a v12 release. We can then bring some of this work (refactored) along for the ride then.

Until then, you should consider a fork of the library for encrypted SQLite instead.

LLM reviewed before create PR 😅
I leave it in 10, but the real problem came with .net 11, we can talk about it future