1.16.0 testunit fails #6559

nunotexbsd · 2024-10-07T11:35:51Z

testunit from previous version 1.15.4 is OK.

> sessionInfo()
R version 4.4.1 (2024-06-14)
Platform: amd64-portbld-freebsd15.0
Running under: FreeBSD 15amd64-main 15.0-CURRENT FreeBSD 15.0-CURRENT 1500023 amd64

Matrix products: default
LAPACK: /usr/local/lib/R/lib/libRlapack.so;  LAPACK version 3.12.0

locale:
[1] C.UTF-8/C.UTF-8/C.UTF-8/C/C.UTF-8/C.UTF-8

time zone: Europe/Lisbon
tzcode source: system (FreeBSD)

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

loaded via a namespace (and not attached):
[1] compiler_4.4.1

<snip>
* checking examples ... OK
* checking for unstated dependencies in 'tests' ... OK
* checking tests ...
  Running 'S4.R'
  Running 'autoprint.R'
  Comparing 'autoprint.Rout' to 'autoprint.Rout.save' ... OK
  Running 'froll.R'
  Running 'knitr.R'
  Comparing 'knitr.Rout' to 'knitr.Rout.save' ... OK
  Running 'main.R'
 ERROR
Running the tests in 'tests/main.R' failed.
Last 13 lines of output:
     0.000s ( 24%) Rereading 1 columns due to out-of-sample type exceptions
     0.000s        Total
  Column 1 <<A>> bumped from 'int32' to 'int64' due to <<123456789123456>> on row 114
            A           B
  "integer64"   "integer"
  Test 2266 produced 1 errors but expected 0
  Expected:
  Observed: no documentation of type 'a' and topic 'o' (or error in processing help)
  Error in eval(exprs[i], envir) : object 'tmp2' not found

  Mon Oct  7 12:28:45 2024  endian==little, sizeof(long double)==16, longdouble.digits==64, sizeof(pointer)==8, TZ==unset, Sys.timezone()=='Europe/Lisbon', Sys.getlocale()=='C', l10n_info()=='MBCS=FALSE; UTF-8=FALSE; Latin-1=FALSE; codeset=US-ASCII', getDTthreads()=='This installation of data.table has not been compiled with OpenMP support.; omp_get_num_procs()==1; R_DATATABLE_NUM_PROCS_PERCENT==unset (default 50); R_DATATABLE_NUM_THREADS==unset; R_DATATABLE_THROTTLE==unset (default 1024); omp_get_thread_limit()==1; omp_get_max_threads()==1; OMP_THREAD_LIMIT==unset; OMP_NUM_THREADS==unset; RestoreAfterFork==true; data.table is using 1 threads with throttle==1024. See ?setDTthreads.', .libPaths()=='/wrkdirs/usr/ports/devel/R-cran-data.table/work/data.table.Rcheck','/usr/local/lib/R/library', zlibVersion()==1.3.1 ZLIB_VERSION==1.3.1
  Error in test.data.table() :
    Failed in 20.2s elapsed (19.4s cpu) after test 2273.3 before the next test() call in /wrkdirs/usr/ports/devel/R-cran-data.table/work/data.table.Rcheck/data.table/tests/tests.Rraw.bz2
  Calls: test.data.table -> stopf -> raise_condition -> signal
  Execution halted
* checking package vignettes ... NOTE
Package has 'vignettes' subdirectory but apparently no vignettes.
Perhaps the 'VignetteBuilder' information is missing from the
DESCRIPTION file?
* DONE

Status: 1 ERROR, 1 WARNING, 3 NOTEs
See
  '/wrkdirs/usr/ports/devel/R-cran-data.table/work/data.table.Rcheck/00check.log'
for details.

*** Error code 1

Any clues?
Thanks

The text was updated successfully, but these errors were encountered:

ben-schwen · 2024-10-07T19:07:22Z

Thanks for the report. Could you check whether the error still exists with the dev version? Updating to dev can be done with data.table::update_dev_pkg()

And could you also post the output of

native_ano = iconv("a\U00F1o", "UTF-8", "")
str(native_ano)
.Internal(inspect(native_ano))

MichaelChirico · 2024-10-07T19:24:39Z

Based on this:

Failed in 20.2s elapsed (19.4s cpu) after test 2273.3 before the next test() call in /wrkdirs/usr/ports/devel/R-cran-data.table/work/data.table.Rcheck/data.table/tests/tests.Rraw.bz2

I guess it's b73d4e9 (i.e., fixed in dev)

You might be able to fix it without updating to dev {data.table} by installing {R.utils} and re-trying.

nunotexbsd · 2024-10-07T19:36:49Z

@ben-schwen

> native_ano = iconv("a\U00F1o", "UTF-8", "")
str(native_ano)
.Internal(inspect(native_ano))
 chr "año"
@2160dad18d58 16 STRSXP g0c1 [REF(1)] (len=1, tl=0)
  @2160d9cc5270 09 CHARSXP g0c1 [REF(3),gp=0x28] [UTF8] [cached] "año"

nunotexbsd · 2024-10-07T19:39:14Z

@MichaelChirico

I have added R.utils to test depends but error is same.
I will try devel and put results here.

BTW, I'm maintaining FreeBSD port: https://www.freshports.org/devel/R-cran-data.table/

nunotexbsd · 2024-10-08T08:22:19Z

Ok, I think I found the problem. OpenMP detection not working as in macports:
https://github.com/macports/macports-ports/blob/99af437d1f41eb15874f926020fdcf39356a7ab2/R/R-data.table/Portfile#L23

And only now I did found it on my build logs:

<...>
===>   Generating temporary packing list
* installing *source* package 'data.table' ...
** package 'data.table' successfully unpacked and MD5 sums checked
** using staged installation
zlib 1.3.1 is available ok
***
*** OpenMP not supported! data.table uses OpenMP to automatically
***   parallelize operations like sorting, grouping, file reading, etc.
*** For details on how to install the necessary toolchains on your OS see:
***   https://github.com/Rdatatable/data.table/wiki/Installation
*** Continuing installation without OpenMP support...
***
** libs
<...>

and included in above unittest log:
getDTthreads()=='This installation of data.table has not been compiled with OpenMP support.

I will try to enable OpenMP and check tests.

nunotexbsd · 2024-10-08T09:13:33Z

I will close this PR as problem is deeper that I thought. I will take care of OpenMP support first and the check further tests.

A quick patch shows that I can build package with OpenMP:

--- configure.orig      2024-10-08 08:42:43 UTC
+++ configure
@@ -144,7 +144,17 @@ detect_openmp () {

   fi # uname=Darwin

-  # No support for OpenMP available
+
+  if [ "$(uname)" = "FreeBSD" ]; then
+
+    export PKG_CFLAGS="${PKG_CFLAGS}"
+    export PKG_LIBS="${PKG_LIBS}"
+    export R_OPENMP_ENABLED=1
+    return
+
+  fi # uname=FreeBSD
+
+ # No support for OpenMP available
   export R_OPENMP_ENABLED=0
 }

But it needs better testing.

Thanks all for help.

….16.0" OpenMP detection broken. See also: Rdatatable/data.table#6559 Rdatatable/data.table@096b20f This reverts commit 5f823f1.

aitap · 2024-10-08T16:03:11Z

locale:
[1] C.UTF-8/C.UTF-8/C.UTF-8/C/C.UTF-8/C.UTF-8

Test 2266 produced 1 errors but expected 0
  Expected:
  Observed: no documentation of type 'a' and topic 'o' (or error in processing help)

Is there a chance that your unit test environment is running with LC_ALL=C instead of LC_ALL=C.UTF-8? It looks like on FreeBSD with LC_ALL=C in action, iconv("a\U00F1o", "UTF-8", "") returns "a?o" (which gets parsed as a help request) instead of NA. iconv() just silently succeeds and returns 1 ("one character converted in a nonreversible way").

For a quick workaround, skip the test if identical(native_ano, 'a?o'). Not sure if R could handle this quirk of FreeBSD's iconv() implementation, but they perform incredible gymnastics to accomodate macOS iconv().

Edit: looks like one of the macOS problems is the same (cf. R_MACOS_LIBICONV_UNDO_TRANSLITERATION). May be worth raising on R-devel to make sure this will be enabled on FreeBSD.

nunotexbsd · 2024-10-09T14:08:37Z

Re-opening since there is nor relation with OpenMP support.

nunotexbsd · 2024-10-11T15:22:49Z

I've managed to get 1.16.2 configure to properly detect openmp:

--- configure.orig      2024-10-11 07:37:20 UTC
+++ configure
@@ -84,7 +84,7 @@ detect_openmp () {

 detect_openmp () {

-  if [ "$(uname)" = "Linux" ]; then
+  if [ "$(uname)" = "Linux" ] || [ "$(uname)" = "FreeBSD" ]; then

     printf "%s" "* checking if R installation supports OpenMP without any extra hints... "
     if "${R_HOME}/bin/R" CMD SHLIB test-omp.c >> config.log 2>&1; then

testunit gives similar results:

<snip>
* checking tests ...
  Running 'S4.R'
  Running 'autoprint.R'
  Comparing 'autoprint.Rout' to 'autoprint.Rout.save' ... OK
  Running 'froll.R'
  Running 'knitr.R'
  Comparing 'knitr.Rout' to 'knitr.Rout.save' ... OK
  Running 'main.R'
 ERROR
Running the tests in 'tests/main.R' failed.
Last 13 lines of output:
    omp_get_max_threads()          12
    OMP_THREAD_LIMIT               unset
    OMP_NUM_THREADS                unset
    RestoreAfterFork               true
    data.table is using 6 threads with throttle==1024. See ?setDTthreads.
  test.data.table() running: /wrkdirs/usr/ports/devel/R-cran-data.table/work/data.table.Rcheck/data.table/tests/tests.Rraw.bz2
  Test 2266 produced 1 errors but expected 0
  Expected:
  Observed: no documentation of type 'a' and topic 'o' (or error in processing help)

  Fri Oct 11 16:16:39 2024  endian==little, sizeof(long double)==16, longdouble.digits==64, sizeof(pointer)==8, TZ==unset, Sys.timezone()=='Europe/Lisbon', Sys.getlocale()=='C', l10n_info()=='MBCS=FALSE; UTF-8=FALSE; Latin-1=FALSE; codeset=US-ASCII', getDTthreads()=='OpenMP version (_OPENMP)==202011; omp_get_num_procs()==12; R_DATATABLE_NUM_PROCS_PERCENT==unset (default 50); R_DATATABLE_NUM_THREADS==unset; R_DATATABLE_THROTTLE==unset (default 1024); omp_get_thread_limit()==2147483647; omp_get_max_threads()==12; OMP_THREAD_LIMIT==unset; OMP_NUM_THREADS==unset; RestoreAfterFork==true; data.table is using 6 threads with throttle==1024. See ?setDTthreads.', .libPaths()=='/wrkdirs/usr/ports/devel/R-cran-data.table/work/data.table.Rcheck','/usr/local/lib/R/library', zlibVersion()==1.3.1 ZLIB_VERSION==1.3.1
  Error in test.data.table() :
    1 error(s) out of 11382. Search tests/tests.Rraw.bz2 for test number(s) 2266. Duration: 29.0s elapsed (00:01:19 cpu).
  Calls: test.data.table -> stopf -> raise_condition -> signal
  Execution halted
* checking package vignettes ... NOTE
Package has 'vignettes' subdirectory but apparently no vignettes.
Perhaps the 'VignetteBuilder' information is missing from the
DESCRIPTION file?
* DONE

Status: 1 ERROR, 1 WARNING, 4 NOTEs
See
  '/wrkdirs/usr/ports/devel/R-cran-data.table/work/data.table.Rcheck/00check.log'
for details.

*** Error code 1

Can't find a way to skip this test.
Any hints?

For a quick workaround, skip the test if identical(native_ano, 'a?o'). Not sure if R could handle this quirk of
FreeBSD's iconv() implementation, but they perform incredible gymnastics to accomodate macOS iconv().

Thanks

aitap · 2024-10-11T15:32:38Z

Can't find a way to skip this test.
Any hints?

This is a bug, should be fixed by #6567. Does it help if you apply it locally?

nunotexbsd · 2024-10-11T15:37:28Z

Can't find a way to skip this test.
Any hints?

This is a bug, should be fixed by #6567. Does it help if you apply it locally?

I can confirm that it fixes test.
Thanks

tdhock · 2024-10-17T00:35:21Z

hi @nunotexbsd thanks for your detailed analysis.
if I understand correctly, there were two issues on freebsd.

the original was about iconv, which was fixed by #6567 so I wonder if this issue can be closed now?

the other was about openmp, for which you proposed a fix #6559 (comment) if that is right then can you please submit a PR?

tdhock · 2024-10-17T00:37:15Z

related to other openmp pr #6497

nunotexbsd · 2024-10-17T07:22:19Z

Hello @tdhock

You can close #6567 as it fixes problem.

Related to #6559 (comment), I will use a local patch for the time being to be sure that it is the better solution for configure phase and then I will open a PR with a patch.

Thanks

nunotexbsd closed this as completed Oct 8, 2024

nunotexbsd reopened this Oct 9, 2024

This was referenced Oct 10, 2024

Non-ascii symbols on systems without UTF-8 #6339

Closed

tests: Detect session encoding incapable of U+00F1 #6567

Merged

MichaelChirico mentioned this issue Oct 11, 2024

Improve by=<non-ASCII> test to be more robust #6573

Open

tdhock added the openmp label Oct 17, 2024

tdhock closed this as completed Oct 17, 2024

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

1.16.0 testunit fails #6559

1.16.0 testunit fails #6559

nunotexbsd commented Oct 7, 2024

ben-schwen commented Oct 7, 2024 •

edited

Loading

MichaelChirico commented Oct 7, 2024 •

edited

Loading

nunotexbsd commented Oct 7, 2024

nunotexbsd commented Oct 7, 2024

nunotexbsd commented Oct 8, 2024

nunotexbsd commented Oct 8, 2024 •

edited

Loading

aitap commented Oct 8, 2024 •

edited

Loading

nunotexbsd commented Oct 9, 2024

nunotexbsd commented Oct 11, 2024

aitap commented Oct 11, 2024

nunotexbsd commented Oct 11, 2024

tdhock commented Oct 17, 2024

tdhock commented Oct 17, 2024

nunotexbsd commented Oct 17, 2024 •

edited

Loading

1.16.0 testunit fails #6559

1.16.0 testunit fails #6559

Comments

nunotexbsd commented Oct 7, 2024

ben-schwen commented Oct 7, 2024 • edited Loading

MichaelChirico commented Oct 7, 2024 • edited Loading

nunotexbsd commented Oct 7, 2024

nunotexbsd commented Oct 7, 2024

nunotexbsd commented Oct 8, 2024

nunotexbsd commented Oct 8, 2024 • edited Loading

aitap commented Oct 8, 2024 • edited Loading

nunotexbsd commented Oct 9, 2024

nunotexbsd commented Oct 11, 2024

aitap commented Oct 11, 2024

nunotexbsd commented Oct 11, 2024

tdhock commented Oct 17, 2024

tdhock commented Oct 17, 2024

nunotexbsd commented Oct 17, 2024 • edited Loading

ben-schwen commented Oct 7, 2024 •

edited

Loading

MichaelChirico commented Oct 7, 2024 •

edited

Loading

nunotexbsd commented Oct 8, 2024 •

edited

Loading

aitap commented Oct 8, 2024 •

edited

Loading

nunotexbsd commented Oct 17, 2024 •

edited

Loading