assigning using := still prints while debugging via browser()

[romunov]

As per FAQ 2.21 and 2.22, printing of DT when using := has been suppressed since 1.8.3. However, I noticed that this still happens in borderline case, e.g. when debugging via browser(). Notice that this happens when one manually runs the assignment line, and not through pressing c, n or Enter. I can reproduce this in Rgui and RStudio (some nightly build).

I can confirm the same behavior in 1.10.5.

> library(data.table)
data.table 1.10.4
  The fastest way to learn (by data.table authors): https://www.datacamp.com/courses/data-analysis-the-data-table-way
  Documentation: ?data.table, example(data.table) and browseVignettes("data.table")
  Release notes, videos and slides: http://r-datatable.com
DT <- data.table(a = 1:3, b = letters[1:3])
DT[, c := rnorm(3)] # doesn't print object

# is browser() somehow responsible?
browser()
Called from: top level 
Browse[1]> DT[, d := rnorm(3)] # doesn't print object
Browse[1]> Q # doesn't appear so
 
sapply(1:3, function(x) {
  DT <- data.table(a = 1:3, b = letters[1:3])
  browser()
  DT[, e := rnorm(3)] # run this manually
})
Called from: FUN(X[[i]], ...)
Browse[1]> DT[, e := rnorm(3)] # prints object
   a b          c          d          e
1: 1 a -0.9089834 -0.8720443 -1.0034940
2: 2 b  1.2618214 -0.9993842  0.3317130
3: 3 c -2.2171694  0.2251432 -0.2082159

Behavior is consistent in lapply as well but not in a for loop.

for(i in 1:3) {
  DT <- data.table(a = 1:3, b = letters[1:3])
  browser()
  DT[, e := rnorm(3)]
}

Called from: top level 
Browse[1]> DT[, e := rnorm(3)] # doesn't print
Browse[1]> 

This is not big deal, I rather like it because I'm curious about my objects when debugging, just something I thought I'd let you know.

> sessionInfo()
R version 3.3.3 (2017-03-06)
Platform: x86_64-w64-mingw32/x64 (64-bit)
Running under: Windows >= 8 x64 (build 9200)

locale:
[1] LC_COLLATE=Slovenian_Slovenia.1250  LC_CTYPE=Slovenian_Slovenia.1250   
[3] LC_MONETARY=Slovenian_Slovenia.1250 LC_NUMERIC=C                       
[5] LC_TIME=Slovenian_Slovenia.1250    

attached base packages:
[1] stats     graphics  grDevices utils     datasets  methods   base     

other attached packages:
[1] data.table_1.10.4

loaded via a namespace (and not attached):
[1] tools_3.3.3 yaml_2.1.14

[MichaelChirico]

The reason it prints from browser() at the top level is that the call stack is small:

[[1]]
(function (x, ...) 
UseMethod("print"))(x)

[[2]]
print.data.table(x)

Whereas in the lapply scenario we've got:

[[1]]
sapply(1:3, function(x) {
    DT <- data.table(a = 1:3, b = letters[1:3])
    browser()
    DT[, `:=`(e, rnorm(3))]
})

[[2]]
lapply(X = X, FUN = FUN, ...)

[[3]]
FUN(X[[i]], ...)

[[4]]
DT[, e := rnorm(3)]

[[5]]
print.data.table(x)

Which doesn't match any of the suppression rules. Whereas without browser() print is not invoked from within the loop:

debug(data.table:::print.data.table)
sapply(1:3, function(x) {
  DT <- data.table(a = 1:3, b = letters[1:3])
  DT[, e := rnorm(3)] # no call to print
})

I'm not sure how worth the effort it is to try and come up w a rule for this situation, as I don't know if it's possible to detect being inside browser().