1352 functions in data raise warnings

[chlebowa]

Fixes #1352

One warning is removed by extending the defunction call in create_app_id to the data argument.
The other is removed by applying defunction to the contents of data when calculating hashes for the integrity check.

Improved recursion condition in function defunction.

[chlebowa]

Tests pass locally now.

[m7pr]

I see 58 issues in tests on local when I run tests for below dependencies. Those are the same as appear in GitHub Actions

devtools::test_file('tests/testthat/test-module_teal.R')

> packageVersion('teal.slice')
[1] '0.5.1.9011'
> packageVersion('teal.data')
[1] '0.6.0.9010'
> packageVersion('teal') #teal comes from this fork
[1] '0.15.2.9066'

[m7pr]

LGTM

[gogonzo]

defunction is a private function. It can't be exposed in show R code!

[chlebowa]

defunction is a private function. It can't be exposed in show R code!

It is not:

foo <- function() cat("hi\n")
mtcars <- mtcars
stopifnot(rlang::hash(foo) == "f5343390f4f30d7df3658452ee946a23")
stopifnot(rlang::hash(mtcars) == "d0487363db4e6cc64fdb740cb6617fc0")
.raw_data <- list2env(list(foo = foo, mtcars = mtcars))
lockEnvironment(.raw_data)

[gogonzo]

defunction is a private function. It can't be exposed in show R code!

It is not:

foo <- function() cat("hi\n")
mtcars <- mtcars
stopifnot(rlang::hash(foo) == "f5343390f4f30d7df3658452ee946a23")
stopifnot(rlang::hash(mtcars) == "d0487363db4e6cc64fdb740cb6617fc0")
.raw_data <- list2env(list(foo = foo, mtcars = mtcars))
lockEnvironment(.raw_data)

Yup. Is the hash in stopifnot matches rlang::hash? Because rlang::hash in show-r-code is evaluated on the foo while in a .get_hashes_code it is on defunction(foo).

Edit:

Running reproducible code fails in the line with rlang::hash(foo) - as stated above. There is a mismatch between provided hash from teal and this generated in src

[chlebowa]

You're right, I didn't actually run the repro code 🤦

[gogonzo]

Closing in favour of

• Accept functions teal.slice#624
• Accept functions #1393