nvimcom:::source.and.clean inconsistently needs plot()

[aavogt]

If I send the png / ggplot / dev.off() lines with <localleader>l, it correctly makes the h.png. If I send the whole file I don't get the h.png (ie. "h.png: cannot open `h.png' (No such file or directory)"). If I add plot( ggplot ... ) or remove some of the apparently unnecessary lines, I get the h.png.

library(tidyverse)
library(ggplot2)

# df <- Sys.glob("*.csv") %>%
#   map_dfr(~ {
#     cbind(f = ., read_csv(., col_names = c("T", "h")))
#   })
# 
# # cat(deparse(dir()))
# ns <- c("10kPa.csv" = 10, "150kPa.csv" = 150, "18e-3kPa.csv" = 18e-3, "220kPa.csv" = 220, "80.5kPa.csv" = 80.5)
# df$p <- ns[df$f]
# write_csv(df, "df.csv")
df <- read_csv("df.csv")

png(file="h.png", width = 400, height = 200)
ggplot(df, aes(p, h, col = T)) +
  ylab("h / (W/m²K)") +
  xlab("p / kPa") +
  geom_point() +
  theme_minimal() + 
  ggtitle("h = a * p^b")
dev.off()
print(system("file h.png", intern = T))
system("rm h.png")

df.csv

f,T,h,p
10kPa.csv,41.13252472675314,4.575119307705421,10

[PMassicotte]

Using source all on this works for me (ie. h.png is created)

library(ggplot2)

png(file = "h.png", width = 400, height = 200)
ggplot(mtcars, aes(x = mpg)) +
  geom_histogram()

dev.off()

[jalvesaq]

I confirm the bug. I get the expected output after <LocalLeader>aa with @PMassicotte's code, but not with @aavogt's one. The code of nvimcom:::source.and.clean is:

function (f, ...) {
    on.exit(unlink(f))
    base::source(f, ...)
}

I don't know what might be wrong...

[PMassicotte]

Interesting, removing the comments will make it work:

library(tidyverse)
library(ggplot2)

# df <- Sys.glob("*.csv") %>%
#   map_dfr(~ {
#     cbind(f = ., read_csv(., col_names = c("T", "h")))
#   })
#
# # cat(deparse(dir()))
# ns <- c("10kPa.csv" = 10, "150kPa.csv" = 150, "18e-3kPa.csv" = 18e-3, "220kPa.csv" = 220, "80.5kPa.csv" = 80.5)
# df$p <- ns[df$f]
# write_csv(df, "df.csv")

df <- tribble(
  ~f, ~T, ~h, ~p,
  "10kPa.csv", round(41.13252472675314, 2), round(4.575119307705421, 2), 10
)

png(file = "h.png", width = 400L, height = 200L)
ggplot(df, aes(p, h, col = T)) +
  ylab("h / (W/m²K)") +
  xlab("p / kPa") +
  geom_point() +
  theme_minimal() +
  ggtitle("h = a * p^b")
dev.off()
print(system("file h.png", intern = T))

[PMassicotte]

But this is working

library(tidyverse)
library(ggplot2)

# df <- Sys.glob("*.csv") %>%
#   map_dfr(~ {
#     cbind(f = ., read_csv(., col_names = c("T", "h")))
#   })
#

png(file = "h.png", width = 400L, height = 200L)
ggplot(mtcars, aes(x = mpg)) +
  geom_histogram()
dev.off()

print(system("file h.png", intern = TRUE))
system("rm h.png")

[PMassicotte]

Note the difference in the comment. Looks like a base R issue?

[jalvesaq]

I agree that it seems to be an issue with base::source. The results are identical if we run base::source() manually on the R Console.

[jalvesaq]

We need a minimal example to put in the Known bugs section and close this issue.

[jalvesaq]

@aavogt, could you please try this in your R.nvim config:

    source_args = "print.eval = TRUE",

Your issue seems to be the opposite of #216

[jalvesaq]

From source help:

Note that running code via source differs in a few respects from entering it at the R command line. Since expressions are not executed at the top level, auto-printing is not done. So you will need to include explicit print calls for things you want to be printed (and remember that this includes plotting by lattice, FAQ Q7.22).

It seems that we should set print.eval=TRUE by default as we have done for nvimcom:::Rnvim.source.

[PMassicotte]

It seems that we should set print.eval=TRUE by default as we have done for nvimcom:::Rnvim.source.

Would make sens.

[jalvesaq]

@aavogt, could you try the print_eval branch, please?