Question: Would a PostgreSQL backend ever be feasible as an optional Calibre database

[TIHTCM]

Hi Kovid and team,
First, thank you for the years of work that have gone into Calibre. It remains one of the most impressive open‑source projects I’ve ever used.

I wanted to ask whether there has ever been consideration for adding an optional PostgreSQL backend for Calibre’s metadata database. I’m absolutely not suggesting replacing SQLite, since it’s perfect for the majority of local, single‑user workflows. I’m only asking about the possibility of a second, opt‑in backend for users who run into issues related to large libraries or networked environments.

Several community discussions note that Calibre’s SQLite database should not be stored on a NAS or network share due to locking and corruption risks, and that this is a long‑standing recommendation from the project. These warnings appear throughout community documentation and user reports, where developers and experienced users emphasize avoiding networked drives to prevent database problems. Similarly, blog posts describing NAS or Docker setups point out the same limitation—that Calibre’s SQLite file should not live on network storage because it was never designed for remote filesystem semantics.

Because of that, I wanted to ask whether a PostgreSQL metadata backend could one day be considered. PostgreSQL would theoretically offer:

Safe remote access without the corruption risks inherent in SQLite over SMB/NFS.
True multi‑user concurrency via MVCC.
Better performance for very large libraries, including efficient indexing and full‑text search.
A pathway to more robust NAS or server‑hosted deployments, especially with Docker setups.

This would not need to change any existing workflow for typical users—SQLite could remain the default, and PostgreSQL could be an advanced/optional configuration for power users who specifically want NAS or multi‑device robustness.
I fully understand if this would be a large architectural change or if it’s something that’s out of scope for Calibre’s direction. But before exploring any prototypes or community experiments, I wanted to check whether this is something the core developer(s) see as:

“Not possible”,
“Not desired for project direction”, or
“Possible in theory but a major undertaking.”

I appreciate any insight you’re willing to share, and again, thank you for everything you’ve built. Calibre continues to be the heart of so many personal libraries—including mine.
Warm regards,
TIHTCM

[kovidgoyal]

Nope. You want safe concurrency, use the calibre server. I am simply never going to maintain multiple datbase backends in calibre, not worth the overhead, in any shape or form.

[TIHTCM]

I completely understand. Thanks for the quick reply. :-)