-
Notifications
You must be signed in to change notification settings - Fork 62
Skip tests for CI with SDC_CONFIG_PIPELINE_SDC=False #340
Skip tests for CI with SDC_CONFIG_PIPELINE_SDC=False #340
Conversation
Merge changes from origin repo
Merge changes from origin repo
Merge changes from origin repo
Merge changes from origin repo
Merge changes from origin repo
Merge changes from origin repo
Merge changes from origin repo
Merge changes from origin repo
Merge changes from sdc master
- Expected that ~35% overall tests are failed
Added by mistake
Merge changes from SDC master
Hello @PokhodenkoSA! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:
Comment last updated at 2019-11-27 10:30:57 UTC |
I know nothing about it. Please ask to merge this PR for people with whom you discussed it |
I mean PR #338 - it is PR from @Hardcode84 with porting to Numba master. Yesterday @Hardcode84 said that this PR was not ready to be merged into master yet but it is in master now. As I understand it is better to revert this PR merge. @Hardcode84, is it correct? So I think the merge of PR #338 to master will be reverted. So for my current PR #340 I take a commit placed in history before the commit where PR #338 was merged to master. I was afraid that merging #338 to master breaks master and it influence on tests. |
…ing config_pipeline_hpat_default.
…(), sdc_jit() and jit(). Tests should use self.jit() instead of sdc.jit().
…skip-test-new-style-master
- Move changes from IntelPython#340 here
…inlining (#407) * remove boost.regex dependency * adapt for get_parfor_reductions interface changes * disable tests due dead code_parfor regression * lambda type_infer quickfix * quick fix * fix define sig * old style: fixes for lambda inlining * fix series combine * some work on df.apply * fix rolling * remove commented code * expected failures * style * changed parfor interface * Set tag numba to ceab10620764d; update llvmlite to 0.31 * Use numpy 1.17; remove builds for numba master * Remove llvmlite version specification for run section in numba recipe * Skip failing tests in new pipeline - Move changes from #340 here * Update numba master commit (#399) * fix functions signatures * update numba master commit * idxmin, idxmax, cumsum workarounds * @overload options * Use numba master for builds in public CI * Fix build of Numba on macOS - Add llvm-openmp and intel-openmp to host section in meta.yaml * Skip 12 failing tests for SDC pipeline - test_df_apply - test_df_apply_branch - test_ts_map_date2 - test_agg_multikey_parallel - test_agg_parallel - test_agg_parallel_std - test_agg_parallel_var - test_agg_seq - test_agg_seq_std - test_agg_seq_str - test_agg_seq_var - test_muti_hiframes_node_filter_agg
This PR is based on PR from @Vyacheslav-Smirnov #301 which configures CI for running with SDC_CONFIG_PIPELINE_SDC=False.
This PR introduces convenient decorators
skip_numba_jit
andskip_sdc_jit
for skipping tests and easy search it in code.Inside this decorators implementations you can see in code how to switch them off (for unskip tests during development) by uncommenting one line.
I have based on the commit which is before #338 merged. From discussion I understand that it is not ready yet to be merged. If it is ok with this merge then my PR should be rebased.
This PR removes TestSeriesNumba because I understand that running numba.jit with SDC_CONFIG_PIPELINE_SDC=True has no sense. So number of tests for series is not doubled again.
In conclusion, my and @Vyacheslav-Smirnov modifications both enable CI with SDC_CONFIG_PIPELINE_SDC=False and skips tests for making it green. I hope so.