Skip to content

Conversation

@hageboeck
Copy link
Member

@hageboeck hageboeck commented Apr 4, 2025

  • Since the missing value feature has been added to RDF and TTreeReader, a crash was observed when running RDF with the missing value feature. When changing to another tree, branches didn't didn't get recreated, leaving them in an inconsistent state.
  • Another crash only observed on Mac15 beta seems to have been a line in a test that's dereferencing a nullptr. It doesn't do anything with that value, but it still generated a SIGTRAP.
  • Python tutorials were running in unspecified order, because a string comparison with an empty string in CMake was taking the wrong branch and removing all dependencies.

Note:

@github-actions
Copy link

github-actions bot commented Apr 5, 2025

Test Results

    19 files      19 suites   4d 14h 2m 14s ⏱️
 2 717 tests  2 716 ✅ 0 💤 1 ❌
49 862 runs  49 861 ✅ 0 💤 1 ❌

For more details on these failures, see this check.

Results for commit 2f653e4.

♻️ This comment has been updated with latest results.

@hageboeck hageboeck force-pushed the TTRVA_initSizeBranch branch 3 times, most recently from 2f653e4 to b17085c Compare April 6, 2025 07:05
@hageboeck hageboeck changed the title Try to fix a crash in TTreeReader Fixes for Mac beta Apr 6, 2025
@hageboeck hageboeck requested a review from dpiparo April 6, 2025 09:04
When changing trees while the missing value feature of RDF was in use,
the proxies for branches could not get recreated, since TTreeReader
refuses the registration when fProxiesSet == true.
This lead to the TTreeReaderValue being in a broken state.
When running over multiple trees, the list of missing proxies in
TTreeReader will grow indefinitely. Here, it is refactored to a set,
which also shortens the code that looks up the missing proxies.
In debugging a UB in TTreeReaderValueBase, it would have been easier to
have an explicit note of the expected error messages of that test.
Instead of globally suppressing all errors, the three expected message
were added here.
On mac15 beta, this generated a SIGTRAP, while it was OK on all other
platforms, because the reference was never used.
But retrieving the pointer instead of dereferencing, the problem
seems to go away.
Due to a missing quote, dependencies on Mac15 beta seemed to have gotten
deleted.
@hageboeck hageboeck force-pushed the TTRVA_initSizeBranch branch from 739ae7e to 2149b61 Compare April 6, 2025 09:11
@dpiparo dpiparo marked this pull request as ready for review April 6, 2025 11:07
@dpiparo dpiparo requested review from bellenot and pcanal as code owners April 6, 2025 11:07
@dpiparo dpiparo merged commit ba83674 into root-project:master Apr 6, 2025
22 checks passed
@hageboeck hageboeck deleted the TTRVA_initSizeBranch branch April 6, 2025 18:11
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants