Conflicts:
- README.md: one WGPU_VALIDATION_INDIRECT_CALL row (this branch's wording).
- apps/designcraft/src/gpu.rs: main's only change is a test tweak in the backend-fallback tests, which now live in gpu/backend.rs; kept this branch's file and ported the tweak there (the check depends on the chosen backend, not the candidate count).
- apps/designcraft/src/main.rs: dropped main's inline VALIDATION_INDIRECT_CALL block (#664) in favour of gpu::instance_flags(); kept the adapter selection, restart and first-frame watch, and took main's macOS Dock Quit handling, preserve_documents_on_exit in on_exit, and profile::config_dir/recovery_dir.
- crates/engine/src/recovery.rs: took main's version; its retained recovered checkpoints do what this branch's recovery change did.
- ROADMAP.md: main's file plus this PR's one progress-log line.
package.sh now also builds designcraft-<ver>-linux-<arch>.pkg.tar.zst with nfpm's archlinux packager,
from the same nfpm.yaml. It is named by the bundle id (ai.storyteller.designcraft), provides designcraft and
conflicts with designcraft, designcraft-bin and designcraft-git, so an AUR helper doesn't take it over and ArtCraft
Launcher can tell it from an AUR build. The README lists it with the other Linux downloads.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
.indd and .indt documents open through File › Open, drop, the web inbox,
file.open, file.openBytes, file.openIdml and the CLI. The engine converts
them to IDML in memory with storytold/indd-utils (pinned by rev, MIT OR
Apache-2.0, no dependencies, builds for wasm) and imports the result
through the existing IDML path. The converter's warnings come first in the
open result's `warnings`, before the IDML importer's and the font
warnings. A file the converter can't read is an open error, shown in the
Can't Open the File alert.
Routing: a path or name ending in .indd/.indt, or bytes starting with the
INDD signature (the web inbox passes the name without its extension).
opens_as_document and the desktop and web Open dialogs list indd/indt.
Tests use synthetic bytes only: signature-bearing junk, a zero-filled
.indd and a short .indt all fail as open errors with no document added;
a dropped unreadable .indd raises the open alert.
(cherry picked from commit bb8a6a23fa)
wgpu validates indirect draw and dispatch arguments with a compute shader it
compiles when the device is created. Metal on an Intel Iris Pro (MacBookPro11,3,
macOS 12.7.6) fails to compile it ("Compiler encountered an internal error" in
MTLCompilerService), the device is lost, and the window never opens.
DesignCraft draws and dispatches nothing indirectly, so leave the check out of the
instance flags. WGPU_VALIDATION_INDIRECT_CALL=1 turns it back on. VectorCraft
does the same (storytold/vectorcraft#651).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
A portable copy unzipped into a read-only folder such as Program Files
kept pointing settings and recovery at a DesignCraftData it could not
write, so both were lost without a word. The data folder is now probed
(created, then a file written and removed) and, when that fails, the
platform folders are used as if there were no portable.txt; the log and
the status bar say which folder could not be written.
DESIGNCRAFT_CONFIG_DIR still wins. Adds
a_portable_install_that_cannot_write_its_data_uses_the_platform_paths.
Inside a macOS bundle the executable sits in DesignCraft.app/Contents/MacOS,
a signed folder nobody edits, so the marker and DesignCraftData are looked
for next to the .app, as craft-settings does. Adds
inside_a_macos_bundle_the_marker_sits_next_to_the_app.
A file named portable.txt beside the executable keeps settings, logs,
the GPU start-up record and crash recovery in a DesignCraftData folder
next to it instead of the per-user folders, as PhotoCraft and FilmCraft
do. A folder of that name is not a marker. DESIGNCRAFT_CONFIG_DIR still
wins, empty or not. The Windows portable zip now ships portable.txt;
the MSI and the macOS app bundle don't, so nothing changes there.
Conflict: ROADMAP.md: both progress-log entries, next's newer ones first
Conflict: apps/designcraft/src/gpu.rs: next's side; the branch's test fix targets the startup record tests next replaced
Conflict: apps/designcraft/src/main.rs: next's gpu startup and macOS IME helper; config_dir() dropped for the branch's profile::config_dir(), recovery_dir from profile::recovery_dir()
Conflict: crates/ui-egui/src/canvas.rs: next's guard (menu_open, keyboard_pane already limits keys to the focused pane)
Conflict: crates/ui-egui/src/lib.rs: modules from both sides
- release.yml: new linux-riscv64 job (cross-compiled on ubuntu-24.04 against
ports.ubuntu.com riscv64 libs, tar.gz only, CLI smoke-tested under QEMU),
ported from GridCraft; the draft release job needs it, so SHA256SUMS covers it.
- packaging/linux/package.sh: CROSS_ARCH/CROSS_TARGET/CROSS_COMPILE/EMULATOR
support, as in GridCraft.
- cargo xtask version set 0.6.0; README, hardware parity and ROADMAP log note
the new platform.
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
- Recovery: file.recovery.open moves each entry to the reopened document's uid
(written before the old one is removed), so recovered documents stay on disk
until saved or closed, also when the app restarts on another adapter or
stops before its first recovery save.
- First-frame watch: a late first frame is logged and noted in the status bar.
The adapter counts as hanging only when the UI thread has been stuck outside
the app's code (logic/ui mark entry and exit) with no progress for 60 s while
the window isn't known to be minimized, occluded or unfocused. The watch asks
for a repaint on each look, so an idle UI ends it. VectorCraft restarts on
the late frame alone, which a held-back compositor or a long compose can set
off.
- Software renderers (WARP, llvmpipe) are passed over while another backend is
left; once no hardware adapter of any backend could show the window, the app
starts again with them allowed, once.
- A restart clears gpu.json's backend entry just before exec or spawn and puts
it back when starting again fails. Backends a restart left out are no longer
recorded as failed.
- The watch ends a hung process with TerminateProcess on Windows.
- The skip list travels as the restarted app's --gpu-skip= argument instead of
an environment variable, so processes the app starts don't inherit it.
Ported from VectorCraft's start-up fixes (#502, #545, #651, the display-GPU
choice, #806, #964) and reconciled with the backend fallback of #297:
- The app picks the window's adapter itself (eframe's native_adapter_selector,
apps/designcraft/src/gpu.rs): adapters that can't present never, then
WGPU_ADAPTER_NAME, hardware before software, native backends before OpenGL,
the GPU that drives the primary display (Windows EnumDisplayDevices through
winsafe, Linux /sys/class/drm), the power preference (WGPU_POWER_PREF;
otherwise power saving on Windows and macOS, the system's order elsewhere),
then DX12 before Vulkan. Every adapter and the choice are logged.
- wgpu's indirect-call validation is off (WGPU_VALIDATION_INDIRECT_CALL=1 turns
it back on): its compute shader failed to compile on an Intel Mac and lost
the Metal device (#334, first error).
- run_native runs under catch_unwind. A wgpu error or a panic in wgpu,
egui-wgpu or naga during the first frames, or a first frame that doesn't
reach the screen in 20 s, starts the app again without that adapter
(DESIGNCRAFT_GPU_SKIP; exec on Unix, a new process on Windows). When no
adapter of the start's backends is left, it starts again with the next
backend. Each restart leaves out one more adapter or backend, so they end.
Adapters without PCI ids (Metal) are keyed by name.
- The backend fallback (gpu/backend.rs, gpu.json) stays for driver faults that
take the process down. Windows tries DirectX 12, then OpenGL, then Vulkan,
each in its own instance: Vulkan last because creating its instance loads
every installed Vulkan driver, and OpenGL kept out of the DX12 instance
because creating it crashed in AMD's atio6axx.dll (#167).
- main returns an ExitCode and logs why the app stopped to standard error and
the log file, which is all a Windows start without a console leaves (#273).
README, architecture, gaps, hardware and target-app parity docs and ROADMAP
describe the choice, the restarts and the variables.
Progress docs per craftrules standards/progress-docs.md: target-app parity, gaps, roadmap with alpha gate, readiness by audience (full / mainstream / essentials) with Opus 5.5 hours, and feature, format, hardware, UI and localization parity docs.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Upstream main moved 9e4f694 -> 736acff by 10 squashes: #258, #285, #153, #280, #288, #281,
#176, #299, #298, #297. New to next: #258 (datamerge test), #153 (temporary files and release
inputs), #280 (warichu), #176 (PDF image conversion cache), #297 (graphics backend before the
window). Already on next: #285, #288, #281, #299, #298.
Conflict: ROADMAP.md: next's lines; CJK1 takes main's wording (ruby unfinished, warichu set after the line is broken)
Conflict: crates/compose/src/lib.rs: list labels (#298): next's list_label structure; paragraph start / break moves: next's `go` (with #295's sub-columns) under the maintainer's #285 comment; Cursor: next's start/next_page/next_baseline with tops; line placement: #280's warichu::place right after layout_line, before next's trimmed trailing spaces and the drop cap splice (so neither is stacked and the spaces follow the closed-up end_x); caret(): next's drop-cap branch first, then #280's stacked-row metrics
Conflict: crates/compose/src/tests.rs: next's tests kept; #280's tests auto-merged
Conflict: crates/engine/src/cmd/text.rs: both (next's TypingFormat/typing_format, #280's warichu_cmd)
Conflict: crates/idml/src/tests.rs: next's side; #280's warichu round-trip test auto-merged
Conflict: crates/pdf/src/export.rs: #176's content cache and test counter with next's RasterMode (no Exporter `interpolate` field on next, so the maintainer's `mode.interpolate &= self.interpolate` is not carried); test exporter gets next's drawing_items field
Conflict: crates/ui-egui/src/canvas.rs: #280's compose::highlight_quads (its RTL branch equals next's mixed-direction spans) followed by next's drop-cap highlight
Conflict: crates/ui-egui/src/menus.rs: next's formatting_affects_text test kept
Auto-merge duplicates removed: #288's table-style tests and #298's list test in idml tests.rs,
#288's tests in doc styles.rs and #285's docx test (both files equal next's), #281's dialog
tests in dialogs.rs (equals next's) and its menu_button/menu_max_height/clipped_submenu in
menus.rs. contributors/people.toml takes main's (next had a duplicated
[people.emanuelebonaventura] table from the 9e4f694 merge, which broke the ui-egui build
script's TOML parse).
Maintainer edits carried over: #285 break characters and Next Column through one closure
(next's `go`, comment added); #281 viewport-aware menu_button in the app bar and the Paragraph
menu, per-dialog widths clamped to the window (all already on next); #280 warichu after the
last-line rules (layout_line with ends_para/forced_mid) and rules at each warichu row's
baseline (rule_baseline in render and pdf text_rules); #176 CMYK fallback through
lossless(.., mode) and its updated contract tests; #258 test fixes; #153 and #297 as landed.
This reverts commit 6e8f33c386, reversing
changes made to de5d32c90f.
#168 is closed; upstream main has #297, which picks the backend before the window exists.
Conflict: apps/designcraft/src/main.rs: kept mod logging; dropped #168's graphics module and its call
* Desktop: pick the graphics backend before the window exists; DirectX 12 on Windows
On an AMD hybrid-graphics laptop (Radeon RX 6800M beside an integrated
Radeon, with two versions of amdvlk64.dll loaded into one process) the
app died a few seconds after launch with an access violation inside
AMD's Vulkan driver at the first swapchain present. Nothing reached the
log: a fault inside a driver is a native crash the engine's panic guard
never sees. wgpu's default backend set (PRIMARY | GL) enumerates Vulkan
first, so every start on that machine went down the same path, while
WGPU_BACKEND=dx12 and WGPU_BACKEND=gl ran fine and the headless CPU
renderer was never affected.
The desktop app now chooses the backend before creating the window
(apps/designcraft/src/gpu.rs): DirectX 12 first on Windows, then Vulkan,
then OpenGL; Vulkan then OpenGL on Linux; Metal on macOS. OpenGL stays
in the set as the in-process fallback when the chosen backend has no
adapter. WGPU_BACKEND still overrides everything and turns the fallback
off.
Start-up fallback: gpu.json in the settings directory records the
backend being tried until the third frame begins (the first has been
presented by then) or the app exits normally. A record left behind means
that backend never showed a frame, so the next start skips it, tries the
next candidate and says so in the status bar. The list starts over when
every candidate has failed or the app version changes;
DESIGNCRAFT_NO_PREFS records nothing. The adapter, backend and driver
version are logged at start-up for bug reports.
Verified on the affected machine: a default start runs on DirectX 12; a
simulated crashed DX12 start moved to Vulkan, which faulted for real,
and the start after that came up on OpenGL with the status line. Nine
unit tests cover the chooser and the record; cargo xtask ci passes.
Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
* Merge origin/main; on Windows OpenGL isn't added to the DX12 or Vulkan instance (#167)
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Fable 5.1 <noreply@anthropic.com>
Co-authored-by: Brandon Thomas <echelon@gmail.com>
* Desktop: write log records to standard error and a rotating log file
The desktop app installed no logger, so every log::warn!/log::error! in
the workspace was discarded: the crash guard's panic report, a failed
crash-recovery save, font folders that can't be read. A launch from a
desktop menu has no terminal either, so there was nothing to attach to a
bug report.
apps/designcraft/src/logging.rs (no new dependency):
- standard error plus <settings dir>/logs/designcraft.log (beside
ui.json and prefs.json); each launch rotates the previous log to
designcraft.1.log and .2.log, so the log of a run that crashed
survives the next start; the file is capped at 16 MiB
- records logged before the settings directory is known are buffered
(bounded) and written once the file is attached
- default levels: info for designcraft*, warn for everything else;
RUST_LOG replaces them with env_logger-style directives
- never panics: an unusable directory or a failed write leaves standard
error as the only sink
The logger is installed first in main (before the engine's panic hook)
and the file is opened after argument parsing (--version leaves no
file). The settings directory moves out of prefs_path into config_dir.
A --control or DESIGNCRAFT_CONTROL_PORT value that is not a port is now
logged instead of silently leaving the control channel off. Documented
in README.md (Quick start > Logs).
* Desktop: no eprintln! on the log and panic paths
eprintln! panics when standard error is a broken pipe (a GUI launch on
Windows), and a panic inside the panic hook aborts. The "no log file"
message goes through log::warn! (standard error only by then); the panic
hook's fallback writes to standard error with write_fmt and ignores the
error. Same fix as storytold/vectorcraft#465.
* CLI: run rejects an export option with no export after it (#67)
--page, --scale and --pdf-options apply to the exports that follow them; one
given after the last --export or --all-pages did nothing and said nothing.
The usage text and docs/agents.md say so.
* MCP: document that text offsets and story lengths are UTF-8 bytes (#68)
* MCP: docs name the UI-only commands precisely; app.links works headless (#62)
* README: PDF and EPUB export ship (#60)
InDesign documents and templates (.indd, .indt) open through File › Open,
drop, the web inbox, file.open, file.openBytes, file.openIdml and the CLI.
The engine converts them to IDML in memory with indd-utils
(github.com/aiongg/indd-utils, rev 9cf456d, MIT OR Apache-2.0, no
dependencies, builds for wasm) and imports the result through the existing
IDML path. The converter's warnings come first in the open result's
`warnings`. A file the converter can't read is an open error ("can't
convert the InDesign file: ..."), shown in the Can't Open the File alert.
Routing: a path or name ending in .indd/.indt, or bytes starting with the
INDD signature (the web inbox passes the name without its extension).
opens_as_document and the desktop and web Open dialogs list indd/indt.
Tests use synthetic bytes only: signature-bearing junk, a zero-filled
.indd and a short .indt all fail as open errors with no document added;
a dropped unreadable .indd raises the open alert.
New designcraft-pptx crate (L3) and file.exportPptx command (File ›
Export PowerPoint…, CLI --export x.pptx). Each page becomes a slide of
native objects: shapes with fills, gradients, strokes and effects,
pictures cropped like their frames, native tables, and text boxes.
Text keeps DesignCraft's composition by default: composed lines end in
line breaks, justified spacing becomes character spacing, and leading
becomes percentage line spacing (PowerPoint rounds exact spacing to
whole points; its 100% line is 1.2 em for every font). Threaded
stories split at their frames and wrapped lines get their own boxes.
lineBreaks: reflow writes one box per frame in native columns instead.
TrueType fonts are embedded as Embedded OpenType when their licences
allow. Objects PowerPoint can't represent (placed PDF/EPS, type on a
path, some effects) are rasterised through the existing
rasterize_where path.
Checked in PowerPoint for Mac and LibreOffice: on the sample magazine,
word positions match DesignCraft's PDF to within 0.8 pt horizontally
and 1.6 pt vertically.
- freebsd job: builds packaging/freebsd/package.sh in a FreeBSD 14.3 VM (same image and
packages as freebsd.yml) into designcraft-<ver>-freebsd-x86_64.tar.gz.
- flatpak jobs (x86_64 on ubuntu-24.04, aarch64 on ubuntu-24.04-arm): repackage the Linux
tarball with packaging/linux/flatpak-bundle.sh and a new bundle manifest; runtime and
finish-args shared with the from-source manifest (checked by packaging-lint).
- AppImages embed gh-releases-zsync update information and ship a .zsync beside them.
- version, linux, web, flatpak and freebsd no longer use the release environment (they read
no secrets), so a workflow_dispatch on a branch dry-runs them; the signing jobs and the
draft release stay environment-gated.
- nfpm package description described an image editor; now describes DesignCraft.
- README: Downloads section listing every build.
Co-authored-by: Brandon Thomas <shared2@storyteller.ai>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
The desktop app installed no logger, so every log::warn!/log::error! in
the workspace was discarded: the crash guard's panic report, a failed
crash-recovery save, font folders that can't be read. A launch from a
desktop menu has no terminal either, so there was nothing to attach to a
bug report.
apps/designcraft/src/logging.rs (no new dependency):
- standard error plus <settings dir>/logs/designcraft.log (beside
ui.json and prefs.json); each launch rotates the previous log to
designcraft.1.log and .2.log, so the log of a run that crashed
survives the next start; the file is capped at 16 MiB
- records logged before the settings directory is known are buffered
(bounded) and written once the file is attached
- default levels: info for designcraft*, warn for everything else;
RUST_LOG replaces them with env_logger-style directives
- never panics: an unusable directory or a failed write leaves standard
error as the only sink
The logger is installed first in main (before the engine's panic hook)
and the file is opened after argument parsing (--version leaves no
file). The settings directory moves out of prefs_path into config_dir.
A --control or DESIGNCRAFT_CONTROL_PORT value that is not a port is now
logged instead of silently leaving the control channel off. Documented
in README.md (Quick start > Logs).
Shippori Mincho (8.7 MB) and its licence leave assets/fonts. Japanese fonts now
come from storytold/craft-fonts when built with CRAFT_FONTS_DIR (never a
Cargo.toml dependency): crates/fonts/build.rs embeds its manifest as
designcraft_fonts::CRAFT_FONTS. Native builds embed all four fonts; wasm32
embeds BIZ UDPGothic Regular only (web size budget).
- UI (egui): the Jpan faces at the end of every family, BIZ UDPGothic first
(Bold first in the semibold family).
- Documents: FontDb loads them after the bundled fonts, Mincho first, so they
are the Japanese fallback ahead of system fonts; SVG text gets them too.
- Vertical alternates test runs on every craft-fonts face and skips when
CRAFT_FONTS is empty; new UI no-tofu test and without-craft-fonts tests.
- FreeBSD CI tests with craft-fonts; release.yml builds every platform with
CRAFT_FONTS_DIR + CRAFT_FONTS_REQUIRED=1 at a pinned commit, packages carry
OFL-<family>.txt; NOTICE, ASSETS.md, AGENTS.md/CLAUDE.md, README, ROADMAP.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
aiongg