Ark: `harp_format()` errors on quosures

DavisVaughan commented 1 month ago

I'm not entirely sure how we got here in the first place, but this came from a private beta user https://github.com/posit-dev/positron-beta/discussions/19#discussioncomment-9407240

[R] 2024-05-12T16:18:40.325464000Z [ark-unknown] WARN crates/harp/src/utils.rs:102: `format()` method for <quosure/formula> should return the same number of elements.
[R] 2024-05-12T16:18:40.326057000Z [ark-unknown] WARN crates/ark/src/variables/variable.rs:299: Error while formatting variable: Error evaluating harp_format(~NULL): Unexpected length from `base::format()`.
[R] R backtrace:
[R] 1. safe_evalq(harp_format(~NULL), <environment>)
[R] 2. withCallingHandlers(list(out, NULL), error = handler)
[R] 3. harp_format(~NULL)
[R] 4. format_oo(x, ...)
[R] 5. format_fallback(x, ...)
[R] 6. stop("Unexpected length from `base::format()`.")

[R] 2024-05-12T16:16:16.793140000Z [ark-unknown] WARN crates/harp/src/utils.rs:102: `format()` method for <quosure/formula> should return the same number of elements.
[R] 2024-05-12T16:16:16.793596000Z [ark-unknown] WARN crates/ark/src/variables/variable.rs:299: Error while formatting variable: Error evaluating harp_format(~everything()): Unexpected length from `base::format()`.
[R] R backtrace:
[R] 1. safe_evalq(harp_format(~everything()), <environment>)
[R] 2. withCallingHandlers(list(out, NULL), error = handler)
[R] 3. harp_format(~everything())
[R] 4. format_oo(x, ...)
[R] 5. format_fallback(x, ...)
[R] 6. stop("Unexpected length from `base::format()`.")

juliasilge commented 1 month ago

This looks to be the same as #2801 to me. What do you think?

DavisVaughan commented 1 month ago

Reprex from #2801

library(ggplot2)

df <- data.frame(x = 1:10, y = rnorm(10))
p1 <- ggplot(data = df, mapping = aes(x=x,y=y)) + geom_point() + theme_bw()
p1

lionel- commented 1 month ago

Two actions we could take here:

Add some handling for quosures so they display the expression and maybe the env id
We could downgrade the offending WARN to a TRACE. The fallback handling works well I think so we shouldn't make it seem like there is something to fix necessarily. The backtrace is probably too distracting in user logs. I included one in this message to make it easier to identify which format classes don't conform to our assumptions. We could just include the class vector in the message instead.

DavisVaughan commented 1 month ago

In particular I noticed this when searching for Error in the log text, so removing the word Error would probably help a lot, as it makes it hard to find "real" problems

posit-dev / positron

Ark: `harp_format()` errors on quosures #3115