Last updated on 2026-11-20 02:51:35 CET.
| Flavor | Version | Tinstall | Tcheck | Ttotal | Status | Flags |
|---|---|---|---|---|---|---|
| r-devel-linux-x86_64-debian-clang | 2.0.1 | 1451.86 | 244.33 | 1696.19 | OK | |
| r-devel-linux-x86_64-debian-gcc | 2.0.2 | 716.75 | 209.77 | 926.52 | OK | |
| r-devel-linux-x86_64-fedora-clang | 2.0.2 | 521.00 | 145.01 | 666.01 | ERROR | |
| r-devel-linux-x86_64-fedora-gcc | 2.0.2 | 855.00 | 241.73 | 1096.73 | OK | |
| r-devel-windows-x86_64 | 2.0.2 | 791.00 | 411.00 | 1202.00 | OK | |
| r-patched-linux-x86_64 | 2.0.1 | 1529.21 | 359.19 | 1888.40 | OK | |
| r-release-linux-x86_64 | 2.0.1 | 1486.55 | 357.44 | 1843.99 | OK | |
| r-release-macos-arm64 | 2.0.2 | 177.00 | -75.00 | 102.00 | OK | |
| r-release-macos-x86_64 | 2.0.2 | 1020.00 | 236.00 | 1256.00 | OK | |
| r-release-windows-x86_64 | 2.0.1 | 1522.00 | 484.00 | 2006.00 | OK | |
| r-oldrel-macos-arm64 | 2.0.2 | OK | ||||
| r-oldrel-macos-x86_64 | 2.0.2 | 960.00 | -48.00 | 912.00 | OK | |
| r-oldrel-windows-x86_64 | 2.0.2 | 778.00 | 520.00 | 1298.00 | OK |
Version: 2.0.2
Check: tests
Result: ERROR
Running ‘testthat.R’ [21s/21s]
Running the tests in ‘tests/testthat.R’ failed.
Complete output:
> library(testthat)
> library(TRexSelector)
>
> test_check("TRexSelector")
TRexGVSSelector::computeLambda2: CV_1SE_IEN_CCD returned lambda_2 = 1.72442 as a BOUNDARY pick: it is the last point of the lambda_2 grid (index 99 of 100), so it is the grid boundary, not a CV optimum; the CV curve is flat or still improving there (low SNR, or the grid does not bracket the optimum). The pick is returned unchanged. Remedies: prefer the 1SE rule over MIN, supply a fixed lambda_2, or refine the lambda_2 grid (cv_n_lambda2 / cv_n_lambda); read getLambda2CVDiagnostics().
TRexGVSSelector::computeLambda2: CV_1SE_IEN_CCD returned lambda_2 = 2.66559 as a BOUNDARY pick: it is the last point of the lambda_2 grid (index 99 of 100), so it is the grid boundary, not a CV optimum; the CV curve is flat or still improving there (low SNR, or the grid does not bracket the optimum). The pick is returned unchanged. Remedies: prefer the 1SE rule over MIN, supply a fixed lambda_2, or refine the lambda_2 grid (cv_n_lambda2 / cv_n_lambda); read getLambda2CVDiagnostics().
TRexGVSSelector::computeLambda2: CV_1SE_IEN_CCD returned lambda_2 = 0.128169 as a BOUNDARY pick: the profiled lambda_1 optimum of this lambda_2 sits on the truncation edge of the lambda_1 path (support cap, deviance rule or end of the path), so the tuner would have preferred a smaller lambda_1. The pick is returned unchanged. Remedies: prefer the 1SE rule over MIN, supply a fixed lambda_2, lengthen the lambda_1 path (cv_n_lambda1), or refine the lambda_2 grid (cv_n_lambda2 / cv_n_lambda); read getLambda2CVDiagnostics().
[TCDIENET] [WARNING] solveAtLambda hit the sweep budget (2000); accepting the current iterate (further hits counted silently, see getSweepCapHits()).
Saving _problems/test-trex_gvs-769.R
ScalingMode::NONE: 6 of 6 design columns have L2 norm outside [0.25, 4]. The dummy blocks follow the unit-L2 standard, so a user-managed design should be unit-L2-commensurate (for z-scored data use ScalingMode::ZSCORE, otherwise rescale the columns).
[TLARS] [WARNING] the input matrix X was already centered and unit-norm (every column within 1e-10) before this solver preprocessed it in place; the second transformation moves the data by rounding. If X was pre-scaled, construct the solver with normalize = false and intercept = false (run as constructed); if a previous solver preprocessed it, call restore() on that solver first.
[TLARS] [WARNING] the input matrix D was already centered and unit-norm (every column within 1e-10) before this solver preprocessed it in place; the second transformation moves the data by rounding. If D was pre-scaled, construct the solver with normalize = false and intercept = false (run as constructed); if a previous solver preprocessed it, call restore() on that solver first.
[TOMP] [WARNING] the input matrix X was already centered (every column within 1e-10) before this solver preprocessed it in place; the second transformation moves the data by rounding. If X was pre-scaled, construct the solver with normalize = false and intercept = false (run as constructed); if a previous solver preprocessed it, call restore() on that solver first.
[TOMP] [WARNING] the input matrix D was already centered (every column within 1e-10) before this solver preprocessed it in place; the second transformation moves the data by rounding. If D was pre-scaled, construct the solver with normalize = false and intercept = false (run as constructed); if a previous solver preprocessed it, call restore() on that solver first.
OMP: Warning #96: Cannot form a team with 4 threads, using 2 instead.
OMP: Hint Consider unsetting KMP_DEVICE_THREAD_LIMIT (KMP_ALL_THREADS), KMP_TEAMS_THREAD_LIMIT, and OMP_THREAD_LIMIT (if any are set).
[ FAIL 1 | WARN 0 | SKIP 0 | PASS 2228 ]
══ Failed tests ════════════════════════════════════════════════════════════════
── Failure ('test-trex_gvs.R:769:3'): an unconverged coordinate descent is an error and the design comes back ──
`sel$select()` threw an error with unexpected message.
Expected match: "sweep budget.*before it converged.*cd_max_sweeps"
Actual message: "c++ exception (unknown reason)"
Backtrace:
▆
1. ├─testthat::expect_error(sel$select(), "sweep budget.*before it converged.*cd_max_sweeps") at test-trex_gvs.R:769:3
2. │ └─testthat:::quasi_capture(...)
3. │ ├─testthat (local) .capture(...)
4. │ │ └─base::withCallingHandlers(...)
5. │ └─rlang::eval_bare(quo_get_expr(.quo), quo_get_env(.quo))
6. └─sel$select()
7. └─TRexSelector:::trex_gvs_select(private$ptr)
[ FAIL 1 | WARN 0 | SKIP 0 | PASS 2228 ]
Error:
! Test failures.
Execution halted
Flavor: r-devel-linux-x86_64-fedora-clang