Fix switch_ts to reset state & clean up IRC when switching TS guesses by calvinp0 · Pull Request #866 · ReactionMechanismGenerator/ARC

calvinp0 · 2026-04-11T08:13:18Z

When a TS guess fails validation (e.g., NMD check), switch_ts picks the next guess but previously left stale state behind:

IRC species from the invalidated guess were never cleaned up. delete_all_species_jobs('TS0') only deletes jobs under the TS0 label, but IRC species like IRC_TS0_1 are separate entries in running_jobs/species_dict/etc. These orphaned species
continued running in parallel with the new guess, potentially interfering with job processing.
job_types flags (freq, sp, opt) were never reset. After guess N's freq completed, job_types['freq'] = True carried over to guess N+1, causing the scheduler to skip re-running freq for the new geometry.
convergence was never reset to None.
The old line self.output[label]['geo'] = ... wrote to the wrong dict level (top-level keys instead of self.output[label]['paths']), making it dead code.
Pending pipe batches from the old guess were never discarded.

Copilot

Pull request overview

Fixes Scheduler.switch_ts() to properly reset state when a TS guess is invalidated, preventing stale jobs/flags from affecting the next TS guess.

Changes:

Clean up IRC species spawned from the invalidated TS guess (remove jobs, output, species entries, labels).
Reset per-label output[label]['job_types'] flags and convergence so the next guess re-runs the pipeline.
Discard pending pipe-batch entries for the invalidated TS label/directions.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 4 comments.

File	Description
`arc/scheduler.py`	Adds TS-switch cleanup for IRC species, resets job state, and clears pending pipe batches.
`arc/scheduler_test.py`	Adds a unit test intended to validate the cleanup/reset behavior.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

arc/scheduler.py

arc/scheduler_test.py

Copilot · 2026-04-11T08:18:57Z

arc/scheduler_test.py

+        sched = self.sched1
+        ts_label = 'methylamine'
+        sched.output = dict()
+        sched.initialize_output_dict()
+
+        # Simulate state after a TS guess completes: freq/sp/opt marked done.
+        sched.output[ts_label]['job_types']['opt'] = True
+        sched.output[ts_label]['job_types']['freq'] = True
+        sched.output[ts_label]['job_types']['sp'] = True
+        sched.output[ts_label]['convergence'] = True
+
+        # Simulate IRC species spawned from the TS guess.
+        irc_label_1 = 'IRC_methylamine_1'
+        irc_label_2 = 'IRC_methylamine_2'
+        irc_spc_1 = ARCSpecies(label=irc_label_1, smiles='CN', compute_thermo=False)
+        irc_spc_2 = ARCSpecies(label=irc_label_2, smiles='CN', compute_thermo=False)
+        sched.species_dict[ts_label].irc_label = f'{irc_label_1} {irc_label_2}'
+        sched.species_dict[irc_label_1] = irc_spc_1


switch_ts() is TS-specific (it calls determine_most_likely_ts_conformer(), expects TS paths under output/rxns/..., and repopulates TS checks). This test uses ts_label = 'methylamine', which is initialized as a non-TS species in setUpClass, so the test setup doesn’t represent a real switch_ts scenario and can’t safely be converted to a true integration-style switch_ts() test without introducing a proper TS species fixture.

arc/scheduler_test.py

Copilot

Pull request overview

Copilot reviewed 2 out of 2 changed files in this pull request and generated 1 comment.

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

Copilot · 2026-04-11T09:34:00Z

arc/scheduler_test.py

+        sched = Scheduler(project='test_switch_ts', ess_settings=self.ess_settings,
+                          species_list=[ts_spc],
+                          opt_level=Level(repr=default_levels_of_theory['opt']),
+                          freq_level=Level(repr=default_levels_of_theory['freq']),
+                          sp_level=Level(repr=default_levels_of_theory['sp']),
+                          ts_guess_level=Level(repr=default_levels_of_theory['ts_guesses']),
+                          project_directory=os.path.join(ARC_PATH, 'Projects',
+                                                         'arc_project_for_testing_delete_after_usage4'),


The test deletes the temporary project directory only at the very end. If an assertion fails earlier, the directory can be left behind and pollute subsequent test runs. Consider registering the cleanup via addCleanup()/try-finally (or using tempfile.TemporaryDirectory) so the directory is removed even on failure.

Suggested change

sched = Scheduler(project='test_switch_ts', ess_settings=self.ess_settings,

species_list=[ts_spc],

opt_level=Level(repr=default_levels_of_theory['opt']),

freq_level=Level(repr=default_levels_of_theory['freq']),

sp_level=Level(repr=default_levels_of_theory['sp']),

ts_guess_level=Level(repr=default_levels_of_theory['ts_guesses']),

project_directory=os.path.join(ARC_PATH, 'Projects',

'arc_project_for_testing_delete_after_usage4'),

project_directory = os.path.join(ARC_PATH, 'Projects',

'arc_project_for_testing_delete_after_usage4')

self.addCleanup(shutil.rmtree, project_directory, ignore_errors=True)

sched = Scheduler(project='test_switch_ts', ess_settings=self.ess_settings,

species_list=[ts_spc],

opt_level=Level(repr=default_levels_of_theory['opt']),

freq_level=Level(repr=default_levels_of_theory['freq']),

sp_level=Level(repr=default_levels_of_theory['sp']),

ts_guess_level=Level(repr=default_levels_of_theory['ts_guesses']),

project_directory=project_directory,

codecov · 2026-04-11T11:21:50Z

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 60.29%. Comparing base (960197e) to head (042494c).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #866      +/-   ##
==========================================
+ Coverage   60.09%   60.29%   +0.19%     
==========================================
  Files         102      102              
  Lines       31045    31075      +30     
  Branches     8087     8098      +11     
==========================================
+ Hits        18658    18736      +78     
+ Misses      10071     9999      -72     
- Partials     2316     2340      +24

Flag	Coverage Δ
functionaltests	`60.29% <ø> (+0.19%)`	⬆️
unittests	`60.29% <ø> (+0.19%)`	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

alongd

Looks good, but I think the addition should be inside delete_all_species_jobs()

calvinp0 · 2026-04-11T19:48:25Z

Looks good, but I think the addition should be inside delete_all_species_jobs()
@alongd I made the change, but also in an arc run found another TS bug, so I added it - so please be aware of this:
https://github.com/ReactionMechanismGenerator/ARC/pull/866/changes#diff-556ad66547f14f3f9a3915808a05b59151ee8216b817845d8d716577c0aeb5ebR3557

When a TS guess fails validation (e.g., NMD check), switch_ts picks the next guess but previously left stale state behind: 1. IRC species from the invalidated guess were never cleaned up. delete_all_species_jobs('TS0') only deletes jobs under the TS0 label, but IRC species like IRC_TS0_1 are separate entries in running_jobs/species_dict/etc. These orphaned species continued running in parallel with the new guess, potentially interfering with job processing. 2. job_types flags (freq, sp, opt) were never reset. After guess N's freq completed, job_types['freq'] = True carried over to guess N+1, causing the scheduler to skip re-running freq for the new geometry. 3. convergence was never reset to None. 4. The old line self.output[label]['geo'] = ... wrote to the wrong dict level (top-level keys instead of self.output[label]['paths']), making it dead code. 5. Pending pipe batches from the old guess were never discarded.

calvinp0 requested review from Lilachn91, alongd and Copilot April 11, 2026 08:13

github-actions bot added the Module: Scheduler label Apr 11, 2026

Copilot started reviewing on behalf of calvinp0 April 11, 2026 08:14 View session

Copilot AI reviewed Apr 11, 2026

View reviewed changes

calvinp0 force-pushed the ts_switch_bugs branch from a6cc203 to a160ef7 Compare April 11, 2026 09:25

calvinp0 requested a review from Copilot April 11, 2026 09:26

Copilot AI reviewed Apr 11, 2026

View reviewed changes

calvinp0 force-pushed the ts_switch_bugs branch from a160ef7 to d66ba64 Compare April 11, 2026 12:45

alongd reviewed Apr 11, 2026

View reviewed changes

calvinp0 force-pushed the ts_switch_bugs branch from d66ba64 to ccbfb89 Compare April 11, 2026 19:16

calvinp0 requested a review from alongd April 11, 2026 19:16

calvinp0 force-pushed the ts_switch_bugs branch from ccbfb89 to a267fe7 Compare April 11, 2026 19:46

calvinp0 force-pushed the ts_switch_bugs branch from a267fe7 to 042494c Compare April 12, 2026 10:57

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Fix switch_ts to reset state & clean up IRC when switching TS guesses#866

Fix switch_ts to reset state & clean up IRC when switching TS guesses#866
calvinp0 wants to merge 1 commit intomainfrom
ts_switch_bugs

calvinp0 commented Apr 11, 2026

Uh oh!

Copilot AI left a comment

Uh oh!

Uh oh!

Uh oh!

Copilot AI Apr 11, 2026

Uh oh!

Uh oh!

Copilot AI left a comment

Uh oh!

Copilot AI Apr 11, 2026

Uh oh!

codecov bot commented Apr 11, 2026 •

edited

Loading

Uh oh!

alongd left a comment

Uh oh!

calvinp0 commented Apr 11, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

Conversation

calvinp0 commented Apr 11, 2026

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Reviewed changes

Uh oh!

Uh oh!

Uh oh!

Copilot AI Apr 11, 2026

Choose a reason for hiding this comment

Uh oh!

Uh oh!

Copilot AI left a comment

Choose a reason for hiding this comment

Pull request overview

Uh oh!

Copilot AI Apr 11, 2026

Choose a reason for hiding this comment

Uh oh!

codecov bot commented Apr 11, 2026 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Codecov Report

Uh oh!

alongd left a comment

Choose a reason for hiding this comment

Uh oh!

calvinp0 commented Apr 11, 2026

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

3 participants

codecov bot commented Apr 11, 2026 •

edited

Loading