Default to utils::view() when no asJSON S3 method exists

[mkoohafkan]

R tibble() objects can contain objects that do not have asJSON methods (e.g., difftime objects, units objects, etc.) . In these cases R will return an error when calling the View() function provided by VSCode-R:

Error: No method asJSON S3 class: units

However, the default utils::View() function can still display these datatypes. It would be great if either:

1. VSCode-R View() function could automatically default to using utils::View() if an asJSON error is encountered, or
2. The error message returned by VSCode-R View() function could direct users to try using utils::View() to display the dataset.

[renkun-ken]

Thanks for reporting! I could reproduce this with the following code:

df <- data.frame(
  diff = difftime(Sys.time() + 1:2, Sys.time(), units = "s")
)
View(df)

Error: No method asJSON S3 class: difftime
Backtrace:
1: base::stop(..., call. = FALSE)
2: stop("No method asJSON S3 class: ", class(x))
3: .local(x, ...)
4: FUN(X[[i]], ...)
5: FUN(X[[i]], ...)
6: vapply(x, asJSON, character(nrow(x)), collapse = FALSE, complex = complex, 
7: .local(x, ...)
8: FUN(X[[i]], ...)
9: FUN(X[[i]], ...)
10: vapply(x, asJSON, character(1), na = na, auto_unbox = auto_unbox, 
11: .local(x, ...)
12: asJSON(x, dataframe = dataframe, Date = Date, POSIXt = POSIXt, 
13: asJSON(x, dataframe = dataframe, Date = Date, POSIXt = POSIXt, 
14: jsonlite::toJSON(x, ...)
15: jsonlite::write_json(data, file, auto_unbox = TRUE)
16: View(df)

Does it work for you if we do either of the following

1. Use jsonlite::toJSON(..., force = TRUE) to force converting to JSON so that difftime will become numerics but without " secs" in the formatted string representation.
2. Format the columns that do not implement asJSON method into their string representation.

[mkoohafkan]

Yeah, that should work great! The string representation can be unhelpful for certain objects, they become very long strings that can be irritating to navigate in a table---but I think that's what utils::View() does too, so it certainly would not be a step backwards. I also think a placholder [OBJECT] string would be fine for objects that can't be converted to JSON, users can't expect every R object to be perfectly displayed!

[mkoohafkan]

Any updates on this? After a recent update, it seems like you can no longer call utils::View() to get a window outside of vscode.

[renkun-ken]

Any updates on this? After a recent update, it seems like you can no longer call utils::View() to get a window outside of vscode.

We don't use force=TRUE yet.

#818 replaces utils::View() so that tibble::view() could work properly. To get a window outside of vscode, you need to specify the following settings in vscode to disable the vscode data viewer:

"r.session.viewers.viewColumn": {
  "view": "Disable"
}

[github-actions]

This issue is stale because it has been open for 365 days with no activity.

[mkoohafkan]

Still no solution for viewing tibbles that include a difftime column.

d = tibble(a = difftime(Sys.time(), Sys.time() - 853563))
View(d)
## Error: No method asJSON S3 class: difftime

[renkun-ken]

I guess the only thing we could do is use force=TRUE then.