| Age | Commit message (Collapse) | Author |
|---|
|
Each thumbnail on a preview page (page-N.html) used to be rendered by its
own "template preview" call, paying the full source_template_file cost --
an "env -i bash" invocation -- per thumbnail. With MAXPREVIEWS thumbnails
per page that was N template renders per page just for the grid.
render_full_preview_page now builds the markup for ALL of a page's
thumbnails in bash (build_preview_thumbnail / append_preview_thumbnail)
and emits the whole grid in ONE render via a new previewpage.tmpl that
takes the pre-built HTML through a context_raw "preview_thumbs" field --
the same pattern the stats filter galleries (camera.tmpl) already use.
Per-thumbnail markup is byte-identical to the old preview.tmpl output:
same <a name=... href=...><img class='thumb <anim>' .../></a> structure,
order, HTML escaping and seeded "slow" animation class. Header and footer
stay as their own template calls, so a page now costs ~1 previewpage
render + header/footer instead of N + chrome.
The parallel job-pool integration and failure contract are unchanged:
each preview page is still one background render job. Added the
render_preview_thumbs_html field spec (hn0 dispatch pattern, context_raw
kind), registered previewpage in the validate_template_dir required
templates and in the required-context-vars test expectations, and pointed
the four generation template-failure tests at previewpage.tmpl (generation
no longer renders preview.tmpl). The standalone "template preview" engine
unit tests keep exercising preview.tmpl, which still ships.
Verified byte-identical output: generated the fixture album (including a
spaces/special-char filename) twice with RANDOM_SEED=42 using the parent
commit's bin/shuriken vs the new bin; every .html file is identical.
just test, just shellcheck, just check-generated and git diff --check all
pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Separate the mixed concerns in bootstrap.source.sh per [SRP]:
- logging.source.sh: output_is_quiet, output_is_verbose, log_info,
log_verbose, log_warning (the output/logging concern).
- paths.source.sh: resolve_default_rc_file, resolve_source_root,
resolve_default_template_dir, resolve_default_asset_dir,
template_dir_uses_default, apply_template_dir_default, init_config
(install/source-root/default-dir + rc-file/template-dir resolution).
- bootstrap.source.sh keeps the CLI usage text and the shared
resolve_config_array parser used by the config modules.
Pure code move, verified byte-identical function bodies. LIB_SOURCES now
lists logging.source.sh first, then bootstrap, then paths; matching
source directives added to src/shuriken.sh so --check-sourced sees the
new modules. just test/shellcheck/check-generated and git diff --check
all pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Pure code move: split the ~1248-line stats.source.sh (SRP+SoC) into three
cohesive modules, all functions and STATS_* constants/globals moved verbatim:
- stats-aggregate.source.sh: EXIF parsing, bucketing, _stats_tally,
reset/collect_photo_exif_stats and the STATS_* maps.
- stats-render.source.sh: stats overview page (sections, bars, leaderboard,
render_stats_page, _stats_build_body, _stats_filter_link, background pickers).
- stats-filter-album.source.sh: per-filter mini-albums (render_filter_pages,
gallery/view/thumbnail builders, _stats_enqueue_filter_album).
Updated LIB_SOURCES accordingly. No logic change (verified: all 55 functions and
top-level globals preserved; full suite, shellcheck, check-generated, diff all
pass). The SoC point (HTML built in bash rather than templates) is left for a
follow-up (nn0/qn0).
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Refactor the ~1442-line album.source.sh ([SRP] violation) into three
cohesive modules, leaving album.source.sh a thin coordinator. Pure code
move, no behavior change -- whole functions and their doc comments moved
verbatim; the assembled bin/shuriken contains the same function set.
- image-pipeline.source.sh: ImageMagick derivative generation
(create_photo_derivatives, create_all_photo_derivatives) and photo
asset prep (prepare_generation_photo_assets).
- album-metadata.source.sh: EXIF identify caching, exif tooltip/details
helpers, file counts, shuriken.json metadata, EXIF cache clearing and
the dry-run plan.
- album-render.source.sh: preview/view/details/splash/redirect page
rendering, the parallel render-job orchestration and the
ALBUM_VIEW_PAGE_BY_PHOTO global.
- album.source.sh: thin coordinator (generate, refresh_splash,
generate_stats_pages, site-asset/archive/html-clean helpers).
Add the three modules to LIB_SOURCES (Justfile) and to the
SHURIKEN_LIB_SOURCES markers in src/shuriken.sh, ordered after random
and before album.source.sh / stats.source.sh.
just build, just test, just shellcheck, just check-generated and
git diff --check all pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Replace the hardcoded `case "$kind"` block in
prepare_template_render_vars with a name-based registration/dispatch
pattern. Each render field kind is now implemented by one
prepare_template_render_var__<kind> handler; the core loop resolves the
handler by name (prepare_template_render_var__$kind), verifies it exists
via `declare -F`, calls it with a uniform signature
(out_nameref, context_name, source_name), and reports a config_error for
an unknown kind (no matching handler) -- preserving the previous error
behavior. Adding a new kind now means defining a new handler function
only; the loop never changes.
Handlers cover all existing kinds: context_css, context_html,
context_raw, current_date_html, config_html (keeps its inner source_name
dispatch for HEIGHT/MAXPREVIEWS/TITLE/etc. and the same :- defaults),
original_basepath_is_set, preview_num_next_html, preview_num_prev_html,
and tarball_include. Escaping and defaults are unchanged, so rendered
output is byte-identical (verified by diff -r of a full --generate
album, before vs after).
Add test_template_render_var_dispatch_is_extensible proving every
declared kind resolves to a handler and that a newly defined handler is
dispatched without touching the core loop (OCP). just test, just
shellcheck, just check-generated and git diff --check all pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Only per-photo view/detail pages were backgrounded; preview pages
(page-*.html) and navigation redirects rendered serially in the main
loop, serializing album generation.
- render_full_preview_page assembles a complete preview page (header,
optional prev link, ordered thumbnails, next/footer) in one call so
the per-page appends stay sequential; queue_preview_page_render_job
backgrounds each whole page via the shared template render pool,
throttled to IMAGE_JOBS.
- render_album_pages now materializes the full deterministic page
layout up front (album_page_records) and enqueues both view jobs and
one preview-page job per page into the same pool; per-photo
bookkeeping moved to _album_record_view_photo.
- render_view_redirects enqueues one job per view page
(render_page_view_redirects) and blocks until all redirects exist,
preserving its synchronous contract while parallelizing the work.
- Removed obsolete advance_album_preview_page and
render_album_page_thumbnail.
Parallelism only affects timing: seeded random choices are
context-keyed, so output is byte-identical across runs. A failed
preview or redirect job flips render_failed and makes generation exit
non-zero, matching the view-job failure contract. The IMAGE_JOBS cap
still bounds concurrent renders.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
source_template_file previously built the BASH_ENV context file by piping
"declare -p ...; declare -f; serialize_template_render_context ...;" into a
fresh "bash -euo pipefail" subprocess for every rendered page. That dumped all
~5000 lines of shuriken functions and spawned a subprocess per page just to run
the serializer - a large per-page cost for albums with hundreds-to-thousands of
pages.
Now serialize_template_render_context runs in the current shell with stdout
redirected straight into the context tempfile, followed by an appended
"unset BASH_ENV". The serializer returns its own non-zero status explicitly so
the failure is detected via an "if" status-test (which returns normally through
the RETURN trap and cleans up the partial context file), robust even when
source_template_file runs inside a status-tested "if template ..." call chain
where bash would otherwise suppress an inner errexit abort.
The trap-based cleanup (RETURN plus INT/TERM/HUP re-raising to $BASHPID) is
preserved unchanged. Two serializer test mocks that relied on the removed
"| bash" errexit now return non-zero explicitly. Rendered HTML output is
unchanged; just test, just shellcheck, just check-generated and
git diff --check all pass.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
The generated pages link a favicon.ico that was always the bundled shuriken
favicon (copy_site_favicon hard-copied share/shuriken/assets/favicon.ico). Add a
FAVICON config variable and a --favicon PATH CLI flag: when set, that file is
published as favicon.ico instead of the bundled default; when empty, the bundled
favicon is used as before.
Plumbed through apply_config_defaults, CLI_OPTION_SPEC + override allowlist,
usage, print_config, the action config list and effective-setting log, and
validated (a non-empty FAVICON must be a readable file) before generation.
shuriken.default.conf and the README document it; a test covers a custom favicon,
its appearance in --print-config, and rejection of a missing file.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
The filter mini-album pages all drew their blurred background from the whole
album, so a filter page's background did not relate to its category. Mirror the
main album instead:
- A filter gallery's background is now a random photo from THAT filter's own set
(like the album preview pages pick a random album photo).
- A filter view page's background is the photo it shows (like the album view
pages).
- The stats overview still uses a random album-wide photo.
Adds _stats_pick_background (random pick from a filter's photo list) and drops
the now-unused album-wide background preload from render_filter_pages.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
The dist directory is the swapped-in staging dir, which mktemp -d creates mode
0700. The published album root therefore stayed 0700, so the first `shuriken
--sync` created the remote album directory 0700 and the web server (daemon) could
not read it -- requiring a manual chmod 755 on each mirror.
Relax the staging dir to the umask-default directory mode right after mktemp -d
(what mkdir would have produced), so the dist root matches its subdirectories and
is served/synced with sane permissions. Add a regression test asserting the dist
root mode equals its photos/ subdirectory mode.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
The filter mini-albums put thousands of HTML files directly in the album root.
Reorganise so only the main album lives in DIST_DIR and all stats content goes
under stats/:
stats/index.html - the stats overview (was stats.html)
stats/<pagebase>/index.html - each filter gallery (was <pagebase>.html)
stats/<pagebase>/<index>.html - each filter view page (was <pagebase>--N.html)
Each mini-album is its own directory, so no single directory holds more than one
filter's pages. Links updated accordingly: the overview links to <pagebase>/, the
gallery thumbnails link to sibling <index>.html, view prev/next/gallery are
same-dir, and assets/album-details/back-to-stats resolve via the ../.. backhref.
The header Stats link and camera.tmpl back link point at stats/index.html.
Mini-albums still reuse the shared photos/thumbs/blurs assets and render in
parallel. Tests and the dry-run plan updated for the new layout; README documents
the stats/ subdirectory structure.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
The per-photo `identify -verbose` cache lived in $DIST_DIR/.shuriken-cache, which
is inside the throwaway staging dir and the published dist: a fresh/cleared dist
lost it (forcing a full, slow re-identify of every original) and it got deployed
to web servers.
Move it to a volatile ./cache/exif directory parallel to ./dist (dirname of
DIST_DIR resolves to the working dir in both the staging and direct contexts).
Now an unchanged photo skips identify on every regenerate, the cache survives a
cleared dist, and it is never deployed. --force clears ./cache/exif once up front
(clear_exif_cache) and the run repopulates it, so force still does exactly one
identify per photo. Dropped .shuriken-cache from the staging-copy loop.
Tests and the audit doc updated for the new cache location.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
_stats_random_background scanned the photos directory once per page; with a
filter mini-album per stat bucket that is thousands of scans. Load the sorted
photo list once into a global (STATS_BG_PHOTOS) that render_filter_pages
populates before forking the render jobs, so each background subshell inherits
the cached array and just indexes into it. The seeded pick is unchanged.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Generalises the per-camera mini-albums to every stats category: clicking any
row on the stats page (camera, lens, year, month, aperture, shutter, ISO, focal
length, megapixels, aspect, orientation, format, exposure program, metering,
white balance, flash) now opens a filter mini-album of just the matching photos,
with view pages whose prev/next cycle within the filter.
Aggregation: a single _stats_tally now records, for every tallied bucket, the
bucket's photo list and a unique filename-safe "pagebase" (camera-canon-eos-r5,
iso-400, year-2023, ...) with collision resolution, exposed via STATS_FILTER_*
maps. The old camera-only STATS_CAMERA_SLUGS/PHOTOS/SLUG_OWNERS are gone.
Render: every bar/leaderboard row is wrapped in a link to its bucket's pagebase
via _stats_filter_link. render_filter_pages replaces render_camera_pages and
renders ALL filter mini-albums (gallery + per-photo view pages) reusing
camera.tmpl/cameraview.tmpl. The galleries/view pages reuse the album's shared
photos/thumbs/blurs assets -- only the HTML differs.
Parallelism: filter pages render through the shared job pool, throttled to
IMAGE_JOBS (the same pool the album view pages use). The album render warms every
photo's EXIF cache (via details pages) before the parallel filter render, so the
filter jobs only read the cache.
Tests updated for the new STATS_FILTER_* structures and extended to assert
non-camera filter mini-albums (orientation, ISO) and their stats-page links.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Two fixes to the per-camera pages:
- The camera thumbnail grid is now full-width and reflows with the window like
the main album. Previously it sat inside the 900px-wide stats column, so wide
thumbnails only fit ~2 per row. camera.tmpl now puts the grid in a full-width
sibling block (centered by the body, like the album's direct thumbnail
children); the heading/back-link stay in the narrow stats column.
- Each camera is now a self-contained mini album. Thumbnails link to per-camera
view pages (camera-<slug>--<index>.html) whose prev/next cycle only through
that camera's photos, instead of linking into the shared album view pages
(where next/prev walked the whole album and left the camera). The view pages
reuse the album's look (blurred background, EXIF tooltip, animation) and offer
Gallery / Details (to the album details page) / Direct-link nav. The "--"
index separator can never collide with a gallery name since slugs never
contain "--".
Adds cameraview.tmpl + its render field, _stats_render_camera_{gallery,views}
and the cameraview body builders. Tests updated for the new thumbnail links and
extended to assert the per-camera view pages and their camera-scoped navigation;
required-context-vars test now also covers the camera and cameraview templates.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Three refinements to the stats site:
- Stats and per-camera pages now get a random blurred background image (same
seeded pick the album preview pages use) instead of a plain black page. Adds
_stats_random_background + STATS_BLURS_DIR; falls back to black when no photos
exist (e.g. isolated unit tests).
- Per-camera thumbnails now behave like the main album: they link to each
photo's album view page (navigation, details, EXIF tooltip) and carry the same
seeded animation class, instead of linking to the raw image. render_album_pages
records a photo -> "<page>-<preview>" map (ALBUM_VIEW_PAGE_BY_PHOTO) that the
camera pages read; photos with no recorded view page fall back to the image.
- Spacing: the camera leaderboard rows get padding + a separator line so the long
wrapping camera names are distinguishable, and the stats sections, headings,
bar rows, back-link, and camera grid get moderate extra spacing so nothing
looks cramped.
Tests updated for the new camera-thumbnail markup and extended to assert the
stats background and the album view-page links; camera.tmpl comment corrected.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
The details view already showed a title="" hover summary (camera, aperture,
ISO, shutter, date) on its image; the normal image view did not. Wire the
same tooltip into view.tmpl:
- render_view_page now computes the EXIF tooltip via photo_exif_tooltip_text
(reusing the shared identify cache, so no extra ImageMagick work) and passes
it as the exif_tooltip context var.
- exif_tooltip is now a required render field for the view template too.
- view.tmpl adds the title="" attribute to its <img> when the summary is
non-empty, mirroring details.tmpl (no title when there is no EXIF).
- Tests: extend the EXIF render test to assert the view image carries the same
tooltip, and update the required-context-vars expectation for view.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Review follow-up: collect_dry_run_plan stored plan["stats_page"] but
print_dry_run_plan never rendered it, so --dry-run gave no signal about
the STATS_PAGE toggle or the stats/camera pages generate() would write
(dead assignment, breaking parity with splash_page).
Print a "Stats page: <yes|no>" line and, when enabled, list stats.html
and camera-*.html under "Planned generated files" (camera pages are a
wildcard since dry-run does not run EXIF aggregation). Add a --dry-run
--no-stats test asserting the stats plan is omitted, and extend the
existing dry-run test to assert the stats lines when enabled.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Final integration of the stats-site feature:
- Add STATS_PAGE config (default yes) mirroring SPLASH_PAGE through every
layer: shuriken.default.conf, apply_config_defaults, CLI_OPTION_SPEC
(--stats/--no-stats), the CLI_CONFIG_OVERRIDE_TARGETS allowlist (without
which the flag was silently ignored), usage text, yes/no validation,
print_config, the action config list + effective-setting log, the
generation metadata, and the dry-run plan.
- generate() now calls a gated generate_stats_pages helper after the album
pages (and before archiving, so the stats/camera pages are tarballed):
collect_photo_exif_stats + render_stats_page . . + render_camera_pages . .
- Emit "stats_page" in shuriken.json next to "splash_page".
- Gate the header Stats nav link behind STATS_PAGE via a new
render_stats_page_html config_html field, so it is hidden (no 404) when
stats are disabled.
- Document STATS_PAGE and --stats/--no-stats in the README.
- Tests: full --generate produces stats.html + camera-<slug>.html + the nav
link and stats_page=true; --no-stats suppresses all three and sets
stats_page=false; print-config and metadata expectations updated.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Implement render_camera_pages <html_dir> <backhref> in stats.source.sh: one
camera-<slug>.html per camera in STATS_CAMERA_SLUGS, reusing the aggregation's
collision-resolved slugs and per-camera photo lists so filenames match pm0's
leaderboard links. Each page reuses the shared header/footer chrome plus a new
camera.tmpl that renders an HTML-escaped camera heading, a pre-built thumbnail
grid, and a back-to-stats link.
Thumbnails reuse preview.tmpl's markup style (thumb image from thumbs/) but link
to the full-size image under photos/ rather than a per-photo view page, since the
view-page name is not derivable outside the album pagination loop. The photos and
thumbs dir names are fixed to match generate()'s render_album_pages call.
Cameras are walked in LC_ALL=C label order for reproducible output, and cameras
with no recorded photos are skipped. Adds the camera_name/camera_thumbs render
field specs and a tests/cli.sh case covering per-camera files, slug collisions,
HTML escaping, thumbnail link targets, and determinism.
This provides the function + template + tests only; wiring into generate() and
the STATS_PAGE toggle remain for rm0.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Review follow-up: _stats_keys_by_count_desc sorted the count tie-break
under the caller's locale, so the generated stats page could differ
byte-for-byte across locales/machines. Pin the sort with LC_ALL=C to keep
the static output reproducible, consistent with the project's seeded /
deterministic generation behavior.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Render the aggregated STATS_* counters from the om0 module into a static
stats.html. render_stats_page builds the variable-length body (camera
leaderboard, temporal/exposure/dimension histograms, format and enum
breakdowns) as HTML and hands it to the new stats.tmpl through a raw
context field, wrapping it with the shared header/footer the way
view/details pages do. Bars are pure CSS so the page stays JavaScript-free.
- new template share/templates/default/stats.tmpl (page chrome + body)
- render_stats_page <html_dir> <backhref> [page_name] plus small section
builders in src/lib/stats.source.sh; EXIF-derived labels are HTML-escaped
- new TEMPLATE_RENDER_FIELD_SPECS field render_stats_body_html (context_raw)
and 'stats' added to render_backhref_html's required list
- camera leaderboard entries link to camera-<slug>.html (um0 owns those)
- Stats nav link wired into the shared header bar
- tests cover leaderboard links, counts/percentages, a histogram section,
&/< escaping, and omission of empty categories
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Review follow-up on the stats aggregation module:
- Distinct camera labels that sanitized to the same slug (e.g. two models
differing only in punctuation) shared one STATS_CAMERA_PHOTOS list and
one camera-<slug>.html link, so um0 would render a merged/ambiguous
per-camera page. Add _stats_resolve_camera_slug, which uniquifies
colliding slugs with a numeric suffix and tracks ownership in a new
STATS_SLUG_OWNERS reverse map. It returns via a nameref (not command
substitution) so the global mutation persists in the caller's shell.
- Align the stats ISO fallback order with album.source.sh's tooltip
builder (ISOSpeedRatings -> PhotographicSensitivity -> ISO) so a photo
with multiple ISO tags buckets the same value it displays.
- Document that the format breakdown trusts the file extension.
- Add test_stats_distinct_cameras_get_unique_slugs covering the collision
fix and slug reuse on re-encounter.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Add src/lib/stats.source.sh implementing the aggregation layer for the
stats site feature. collect_photo_exif_stats() iterates the album's
incoming photos, reads each one's cached identify -verbose output via
album.source.sh's cache helper, and fills global STATS_* associative
arrays for the camera leaderboard, per-camera photo lists, temporal,
exposure, dimension, format and enum stats. Builds exactly the v1 stat
set recommended by docs/stats-exif-audit.md (rendering is left to the
sibling tasks pm0/rm0/um0).
Wire stats.source.sh into the Justfile LIB_SOURCES list (after album,
before config) and regenerate bin/shuriken. Add unit tests exercising
synthetic identify fixtures, rational/bucket boundaries, DateTimeOriginal
substring parsing, missing-field tolerance, and the cached-output path.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
The signal handler in source_template_file re-raised the caught signal
with `kill -s $sig $$`. Because the function commonly runs inside
backgrounded render subshells (queue_album_view_render_job), $$ expands
to the main shuriken PID, not the subshell's own. On interrupt the
handler therefore signalled the main shell -- which had already cleared
its own staging traps -- killing it mid-cleanup (leaking the staging
dir) instead of terminating just the render subshell.
Re-raise to $BASHPID, the current (sub)shell's real PID, which equals $$
in the foreground case so it is correct everywhere. Extend
test_template_interrupt_removes_context_file to run the render in a
backgrounded subshell and assert the parent shell survives the re-raise;
verified it fails against a $$-based build and passes with $BASHPID.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
source_template_file created its context tempfile with mktemp but only ran
rm -f on the success and explicit-failure paths. When a render was interrupted
by a signal (terminate_active_generation SIGTERMs the backgrounded render
subtree on Ctrl-C/abort) the tempfile leaked.
Register cleanup in exactly one place, inline in source_template_file's body:
- A RETURN trap covers normal and error returns and clears all of these traps
(including itself) so it cannot linger and fire on an enclosing function's
return against the out-of-scope context_file local (set -u). The trap must be
set in the function body, not a helper: without functrace a RETURN trap is not
function-scoped and would fire when a helper returns, deleting the file before
the render runs.
- INT/TERM/HUP handlers remove the file, clear the traps and re-raise so the
process still exits with the signal's default disposition. PIPE is not
trapped (the internal context-build pipeline emits legitimate SIGPIPE);
SIGKILL is untrappable, leaving only the OS-reaped KILL-escalation case.
The redundant rm -f calls are removed. Adds
test_template_interrupt_removes_context_file, which runs an in-flight render and
SIGTERMs it, asserting the context file is gone.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
The shared resolve_config_array helper returns non-zero when the source
config var was never declared. Both callers (resolve_tar_opts,
resolve_sync_destinations) invoked it unconditionally and ignored the
return value, so under set -euo pipefail a genuinely-unset TAR_OPTS or
SYNC_DESTINATIONS would abort the function before the empty-array
fallback could run. Currently masked because apply_config_defaults
always declares both vars, but the contract was unsafe to rely on.
Guard both call sites with `|| true` and document why: the empty array
the helper leaves behind is exactly what each caller's fallback expects.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
Introduce a shared resolve_config_array helper in bootstrap.source.sh that
parses a config variable declared as either a Bash array or a
whitespace-separated scalar into a named output array. resolve_tar_opts and
resolve_sync_destinations now both delegate to it, with resolve_tar_opts
keeping its "-c" default for the empty/unset case.
validate_imagemagick now reuses resolve_imagemagick_command instead of
duplicating the magick/convert probing, reporting failures through
config_error so the validation output is unchanged.
Add a print-config test covering the empty TAR_OPTS (scalar and array)
fallback to the default.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Task: am0
|
|
|
|
|
|
|
|
|
|
|
|
|
