Closed
Description
@jeroen now builds under arm64 on Linux, and this leads to four Sugar tests failing. From the current log
Status: 1 ERROR
See
‘/__w/rcppcore/rcppcore/Rcpp.Rcheck/00check.log’
for details.
Error: Check: tests, Result: ERROR
Running ‘tinytest.R’ [223s/195s]
Running the tests in ‘tests/tinytest.R’ failed.
Last 13 lines of output:
----- FAILED[data]: test_sugar.R<69--69>
call| expect_true(is.na(fx(c(NA, 6))))
diff| Expected TRUE, got FALSE
----- FAILED[data]: test_sugar.R<77--77>
call| expect_true(is.na(fx(c(5, NA))))
diff| Expected TRUE, got FALSE
----- FAILED[data]: test_sugar.R<86--86>
call| expect_true(is.na(fx(c(NA, 1))))
diff| Expected TRUE, got FALSE
----- FAILED[data]: test_sugar.R<1623--1623>
call| expect_true(is.na(intmin(integer(0))), "min(integer(0))")
diff| Expected TRUE, got FALSE
info| min(integer(0))
Error: 9 out of 1612 tests failed
Execution halted
tt so happens that r2u also has arm64 support so should be easy for me to poke at in CI so will take a look.
Metadata
Metadata
Assignees
Labels
No labels
Activity
jeroen commentedon May 5, 2025
How did you find out so fast :o
eddelbuettel commentedon May 5, 2025
It so happens that I have a fairly clean slate at 'my' r-universe', which lead me to filling in one really old 'pot hole', namely for a never-released package failing on non-amd64 platforms. So I had worked on that last eve, checked this morning and noticed the (new !) 'trigger rebuild' icon for the Rcpp line. So I looked.
Turns out we were already skipping a few NA related tests on arm64-macos so the Rcpp fix should be simple. I'll ping you over DM with another suggestion.