| Commit message (Collapse) | Author | Age | Files | Lines |
|---|
| |
|
|
|
|
|
|
| |
savehist rewrites its whole file on every save, keeping only the variables registered in the running process. Any Emacs that saved savehist without wttrin loaded (a batch tool, a session that never opened wttrin) silently deleted the persisted favorite and saved-location entries. A d-set default then reverted to the init value on the next restart.
The runtime favorite and directory now live in wttrin-state-file, written only by wttrin: write-through on every location command, atomic temp-file + rename, UTF-8 pinned both ways, read with read (never eval), versioned plist. Values restore at load and re-assert on savehist-mode-hook, so a stale legacy savehist restore can't clobber them. With no state file present, legacy savehist values are adopted and written once, so existing users migrate with zero steps. Search history stays on savehist. It's scrub-tolerant throwaway data, exactly what savehist is for.
Also: the README documents the mode-line tooltip forecast option, and the gitignore covers one more local tooling artifact.
|
| |
|
|
|
|
| |
Five test files hand-rolled the same scaffolding to capture the echo-area message: a nil-initialized var, a cl-letf on message, and a lambda that stored the formatted string. Centralize it as testutil-wttrin-with-captured-message, which binds the var and captures the last message shown.
Behavior is unchanged. The full suite stays green. Sites that mock message for a different reason are left alone: the ones that silence it with #'ignore, and the one that accumulates every message into a list.
|
| |
|
|
|
| |
Update copyright headers across all 37 .el files to include 2026.
Add missing Author field to testutil-wttrin.el for consistency.
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
Replace TTL-based cache invalidation with proactive scheduled refresh.
Both mode-line and buffer systems now follow: timer refreshes cache,
display reads from cache, staleness indicated when data is old.
Phase 1 - Mode-line cache formalization + staleness display:
- Replace wttrin--mode-line-tooltip-data with wttrin--mode-line-cache
as (timestamp . data) cons cell matching buffer cache pattern
- Add wttrin--format-age helper for human-readable age strings
- Rewrite wttrin--mode-line-update-display to take no arguments,
read from cache, compute staleness (age > 2x refresh interval),
dim emoji gray when stale, show staleness info in tooltip
- Rewrite wttrin--mode-line-fetch-weather to write cache on success,
show stale display on failure with cache, error placeholder without
- Add wttrin--mode-line-update-placeholder-error for first-launch failure
Phase 2 - Remove TTL, add proactive buffer refresh:
- Rename wttrin-cache-ttl to wttrin-refresh-interval (default 3600s)
with define-obsolete-variable-alias for backward compatibility
- Change wttrin-mode-line-refresh-interval default from 900 to 3600
- Remove TTL check from wttrin--get-cached-or-fetch; serve cached data
regardless of age, background timer keeps it fresh
- Add buffer refresh timer (wttrin--buffer-cache-refresh)
Phase 3 - Buffer staleness display:
- Add wttrin--format-staleness-header for buffer age display
- Insert staleness line in wttrin--display-weather before instructions
Phase 4 - Cleanup:
- Remove all references to wttrin--mode-line-tooltip-data
- Update README.org cache settings and mode-line documentation
- Update tests for new API (198 tests across 21 files, all passing)
|
| |
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| |
- Expand testutil-wttrin.el with shared fixtures and macros
(testutil-wttrin-with-clean-weather-buffer,
testutil-wttrin-mock-http-response, sample ANSI/weather constants)
- Consolidate cache tests: port unique tests from cleanup-cache-constants
and cleanup-cache-refactored into cleanup-cache-if-needed, delete
obsolete files
- Extract startup-delay tests into dedicated file
- Add setup/teardown and (require 'testutil-wttrin) to all test files
- Rename all 160 tests to follow
test-<module>-<category>-<scenario>-<expected-result> convention
- Rename integration test file and add detailed docstrings
- Update Makefile glob to discover new integration test naming pattern
Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
|
|
|
Session 1: Testing infrastructure and initial test coverage
Bug fixes in wttrin.el:
- Fix wttrin-additional-url-params to handle nil unit system
- Remove incorrect callback parameter to url-retrieve-synchronously
- Add nil buffer check for network failures
- Strip HTTP headers before decoding response
- Kill buffer after fetch to prevent memory leaks
- Fix double concatenation of URL params in cache function
- Add proper URL encoding via new wttrin--build-url function
Refactoring:
- Extract wttrin--build-url as pure, testable function
- Separate URL building logic from network I/O
Test infrastructure (33 tests, 100% passing):
- tests/testutil-wttrin.el: Shared test utilities
- tests/test-wttrin-additional-url-params.el: 7 tests
- tests/test-wttrin--make-cache-key.el: 9 tests
- tests/test-wttrin--build-url.el: 10 tests
- tests/test-wttrin--cleanup-cache-if-needed.el: 7 tests
Documentation:
- docs/testing-plan.org: Comprehensive testing roadmap
- docs/bugs.org: Bug analysis from code review
- docs/NOTES.org: Session tracking and guidelines
- docs/session-1-summary.org: Detailed session summary
Next session: Cache workflow tests, parsing logic extraction, integration tests
🤖 Generated with [Claude Code](https://claude.com/claude-code)
Co-Authored-By: Claude <noreply@anthropic.com>
|
