fix: Close Directory and Delete Cache reliability (#27, #34) #112

Merged

ValleyGeek merged 4 commits from 0.1.37/issue-27-34-close-directory into main 2026-06-20 07:41:26 +00:00

Conversation 0 Commits 4 Files changed 7 +185 -39

Claude commented 2026-06-20 07:39:41 +00:00

Member

Copy link

Summary

• #27 — "Close Directory and Delete Cache" reported the cache file as still in use on Windows. Root cause: every WorkingDB/GlobalConfigDB read or write opened a SQLite connection via a bare with self._connect() as conn:, which only commits/rolls back the transaction — it never closes the connection. A local repro confirmed this leaks file descriptors even after an explicit gc.collect() (200 connections → ~100 leaked fds), explaining a Windows handle64.exe dump showing dozens of simultaneous open handles to the working DB and global config DB. Every call site now routes through the existing properly-closing transaction() helper in both DB classes. Kept the earlier WAL-checkpoint + Windows-mmap-disable mitigation as defense in depth.
• #34 — Closing a directory via "Close Directory and Delete Cache" didn't clear it from the recently-opened list, so it kept getting auto-reopened on next launch when "open last directory on startup" is enabled. A _remove_from_recent() helper already existed but was dead code; the close handler now calls it.

Test plan

• ruff check / ruff format --check
• Full pytest suite (372 passed, 1 skipped)
• New regression tests that fail against the old unclosed-connection pattern and pass against the fix (test_repeated_reads_do_not_leak_connection_handles, test_repeated_pref_reads_do_not_leak_connection_handles)
• New test for #34 (test_close_directory_and_delete_cache_clears_last_opened)
• ./scripts/smoke-ui-exit.sh — no Qt shutdown errors
• ./scripts/check-version-bump.sh — VERSION bumped 0.1.37 → 0.1.38
• Confirmed by the reporter against a real Windows build

Closes #27, #34.

## Summary - **#27** — "Close Directory and Delete Cache" reported the cache file as still in use on Windows. Root cause: every `WorkingDB`/`GlobalConfigDB` read or write opened a SQLite connection via a bare `with self._connect() as conn:`, which only commits/rolls back the transaction — it never closes the connection. A local repro confirmed this leaks file descriptors even after an explicit `gc.collect()` (200 connections → ~100 leaked fds), explaining a Windows `handle64.exe` dump showing dozens of simultaneous open handles to the working DB and global config DB. Every call site now routes through the existing properly-closing `transaction()` helper in both DB classes. Kept the earlier WAL-checkpoint + Windows-mmap-disable mitigation as defense in depth. - **#34** — Closing a directory via "Close Directory and Delete Cache" didn't clear it from the recently-opened list, so it kept getting auto-reopened on next launch when "open last directory on startup" is enabled. A `_remove_from_recent()` helper already existed but was dead code; the close handler now calls it. ## Test plan - [x] `ruff check` / `ruff format --check` - [x] Full `pytest` suite (372 passed, 1 skipped) - [x] New regression tests that fail against the old unclosed-connection pattern and pass against the fix (`test_repeated_reads_do_not_leak_connection_handles`, `test_repeated_pref_reads_do_not_leak_connection_handles`) - [x] New test for #34 (`test_close_directory_and_delete_cache_clears_last_opened`) - [x] `./scripts/smoke-ui-exit.sh` — no Qt shutdown errors - [x] `./scripts/check-version-bump.sh` — VERSION bumped 0.1.37 → 0.1.38 - [x] Confirmed by the reporter against a real Windows build Closes #27, #34.

Claude added 4 commits 2026-06-20 07:39:41 +00:00

fix(db): release WAL/mmap before deleting workspace cache (fixes #27) ... 78867cf904

Skip mmap'ing the local working DB on Windows (same reasoning the
share-side pragmas already use) and run a best-effort WAL checkpoint +
journal-mode reset right before "Close Directory and Delete Cache"
attempts to unlink the file, so the cache deletion isn't left racing a
lingering memory-mapped file handle.

fix(ui): clear last-opened directory on Close Directory and Delete Cache (fixes #34) ... 955552c27c

_remove_from_recent() existed but was never called from the close
handler, so the GlobalConfigDB recent_dirs entry (and therefore the
auto-open-last-directory feature) kept pointing at a workspace whose
cache had just been deleted.

fix(db): explicitly close every short-lived connection (fixes #27) ... cc4b1d7d05

sqlite3.Connection's own __enter__/__exit__ only manage the transaction
(commit/rollback) — they never close the connection. Every WorkingDB and
GlobalConfigDB read/write opened a fresh connection via a bare
"with self._connect() as conn:" and relied on CPython refcounting to
release it, which a local repro confirmed does *not* happen reliably:
200 short-lived connections opened that way leaked ~100 file descriptors
even after an explicit gc.collect(), while the same loop using explicit
conn.close() stayed flat. On Windows this leaves dozens of simultaneous
handles to the working DB (and its -wal/-shm companions) and to the
global config.db open for the life of the session — confirmed against a
handle64.exe dump from a real repro — which is what made "Close Directory
and Delete Cache" keep reporting the cache file as still in use well past
the existing 10s retry window.

Both classes already had a properly-closing transaction() context
manager; every bare _connect() call site now routes through it instead.
The earlier WAL-checkpoint/mmap-on-Windows mitigation (previous commit)
stays in place as defense in depth, but this is the actual root cause.

Threading notes: no behavior change for worker threads — each call still
opens and now properly closes its own short-lived connection, consistent
with the existing "connections are short-lived, never shared" contract.

docs: PR-prep for issues 27 and 34 - bump to 0.1.38 bc85674c12

ValleyGeek merged commit 7dd81c3391 into main 2026-06-20 07:41:26 +00:00

ValleyGeek deleted branch 0.1.37/issue-27-34-close-directory 2026-06-20 07:41:26 +00:00

ValleyGeek referenced this pull request from a commit 2026-06-20 07:41:27 +00:00

fix: Close Directory and Delete Cache reliability (#27, #34) (#112)

Sign in to join this conversation.

Reviewers

No reviewers

Labels

Clear labels   

Kind/Bug

Something isn't working as expected

  

Kind/Feature

New capability or user-requested feature

  

Priority/High

High priority

  

Priority/Medium

Medium priority

  

Reviewed/Confirmed

Bug or feature report confirmed valid

  

Compat/Breaking

Breaking change that won't be backward compatible

  

Kind/Bug

Something is not working

  

Kind/Documentation

Documentation changes

  

Kind/Enhancement

Improve existing functionality

  

Kind/Feature

New functionality

  

Kind/Security

This is security issue

  

Kind/Testing

Issue or pull request related to testing

  

Priority

Critical

The priority is critical

  

Priority

High

The priority is high

  

Priority

Low

The priority is low

  

Priority

Medium

The priority is medium

  

Reviewed

Confirmed

Issue has been confirmed

  

Reviewed

Duplicate

This issue or pull request already exists

  

Reviewed

Invalid

Invalid issue

  

Reviewed

Won't Fix

This issue won't be fixed

  

Status

Abandoned

Somebody has started to work on this but abandoned work

  

Status

Blocked

Something is blocking this issue or pull request

  

Status

Need More Info

Feedback is required to reproduce issue or to continue work

No labels

Kind/Bug

Kind/Feature

Priority/High

Priority/Medium

Reviewed/Confirmed

Compat/Breaking

Kind/Bug

Kind/Documentation

Kind/Enhancement

Kind/Feature

Kind/Security

Kind/Testing

Priority

Critical

Priority

High

Priority

Low

Priority

Medium

Reviewed

Confirmed

Reviewed

Duplicate

Reviewed

Invalid

Reviewed

Won't Fix

Status

Abandoned

Status

Blocked

Status

Need More Info

Milestone

Clear milestone

No items

No milestone

Projects

Clear projects

No items

No project

Assignees

Clear assignees

No assignees

1 participant

Notifications

Subscribe

Due date

The due date is invalid or out of range. Please use the format "yyyy-mm-dd".

No due date set.

Dependencies

No dependencies set.

Reference

ai-collab/bulk-image-organizer!112

No description provided.