Access to the local file system

[gregvolny]

Since my last post, a lot of progress has been done with WebR. And, I'm coming back with a new request...
a) Can we use this new API : https://developer.chrome.com/blog/persistent-permissions-for-the-file-system-access-api
for accessing/using and saving files ( for example, downloaded packages) to and from the local file system?

In the meantime, do you have a workaround for accessing to a sqlite database stored in the local file system without any api call?
I tried to upload it to a new directory created in the virtual file system and use

library(DBI)
setwd(file.path("/", "Sqlite","inv_Agr.db3"))
con <- dbConnect(RSQLite::SQLite(),"inv_Agr.db3")

without any success...
it display
Error in setwd(file.path("/", "Sqlite", "inv_Agr.db3")) :
cannot change working directory

[georgestagg]

Can we use this new API [...] for accessing/using and saving files ( for example, downloaded packages) to and from the local file system?

No, not currently. There are deep technical reasons that makes using these kinds of browser APIs difficult in webR. We're tracking this in #56

---

In the meantime, do you have a workaround for accessing to a sqlite database stored in the local file system without any api call?
I tried to upload it to a new directory created in the virtual file system and use [...]

Can you please tell me more about your setup? Are you running webR from the REPL at https://webr.r-wasm.org/latest/, or in your own webR application? How are you uploading the sqlite database to the virtual filesystem? Are you using the webR JavaScript API? Can you create a small self-contained example of your use case for me to experiment with?

---

If you are using your own application there is a possibility for persistent storage, but it requires that webR is loaded using the PostMessage communication channel. When webR is loaded using that channel, you can use the IDBFS filesystem type to mount a directory that will be persisted in the user's web browser:

dir.create("/data")
webr::mount(mountpoint = "/data", type = "IDBFS")
webr::syncfs(TRUE)

Write to the directory by writing to /data and then persisting the data using webr::syncfs(FALSE):

write.csv(mtcars, "/data/mtcars.csv")
webr::syncfs(FALSE)

Then, when the mountpoint is mounted in the future (say, after the page has been refreshed), the data will be available again after running webr::syncfs(TRUE) and the underlying JavaScript promise has resolved.

There are some subtleties here, since the underlying mechanism is asynchronous. Some more documentation is available at: https://docs.r-wasm.org/webr/latest/mounting.html#indexeddb-filesystem-storage

[gregvolny]

Can you please tell me more about your setup? Are you running webR from the REPL at https://webr.r-wasm.org/latest/

I was using https://webr.r-wasm.org/latest/

How are you uploading the sqlite database to the virtual filesystem?

I was using Upload file from the main UI and trying to upload a 261MB Sqlite file. While I was able to upload lighter sqlite, the 261MB one failed.

Are you using the webR JavaScript API? Can you create a small self-contained example of your use case for me to experiment with?

I'm following this project for nearly 2 years. My main goal was to implement dynamic complexes tabulation with data stored on remote servers: PostgreSQL and MySQL. However, until now, I never try a demo since I'm still waiting access to remote database in the API. This could be changed if you implement this: #490