Tilman Roeder
Tilman Roeder
I think this is correct. I’m not sure we can make this work correctly across workers; but we could add some bookkeeping internal to the VFS to address this issue?
1. Yes there is; I think this results in a combination of sleeps / errors returned back to the caller 2. There is a sleep call in the vfs, but...
This also seems to happen very frequently inside of various PPXs.
The signature help RPC also seems to have similar, but less frequently occurring issue.
I think it’s a similar issue to go-to-definition, but it seems to happen less frequently; so addressing the go-to-definition noise is much higher value.
There is a [`C` API](https://github.com/dyedgreen/cqlite/blob/master/cqlite.h), so one could definitely write Python bindings on top of that. (Or maybe even easier, one could build it on top of https://github.com/PyO3/pyo3)
I think the browser version also does something very similar using an import map. I think you could remap the existing WASM module to a different one, to get a...
We could support both relatively easily (assuming we make a VFS that supports OPFS). The current browser support uses an import map to replace the VFS so it’s quite easy...
I’d be happy to have support for something like this! Depending on how it’s done, it might even be possible to make this work with a one-time runtime check at...
This won’t help with performance, but would mean we can open wal mode sqlite files without error so I think enabling this by default would be useful!