1cd58bcaaafc41cbe6670662759a6b291f2919da
Touch has had a range gesture for as long as selection mode has: hold to
enter it, then double-tap the far cell. Nobody finds it. It is invisible,
it is unreliable on a grid that scrolls under the second tap, and it
extends from the anchor as it was *before* the two taps moved it — a rule
subtle enough to need two paragraphs of Rust to explain itself.
The deeper problem is the timer. A double tap is bounded by the double-tap
interval, so the two ends have to be on screen together. The ranges that
actually hurt on a tablet are longer than a screenful, and those are
exactly the ones it cannot describe — which is also why a drag-to-select
sweep would not have fixed this, and why it is not what went in.
So: a "Select to…" button arms the range, the strip stops reporting and
starts instructing ("Tap the last photograph — scroll first if you need
to"), and a Cancel puts it down. Nothing is timing the two taps, so the
user may scroll as far as they like between them, and the run is resolved
by the catalog rather than by what happens to be loaded.
An armed range reaches Rust as `cell-pressed`'s shift argument, because
that is what it is: `apply_press` already reads ctrl+shift as "add the run
from the anchor to here", and in selection mode ctrl is already set. So
this needs no Rust state and no third selection policy — one copy of the
rules, in the function that had them.
"Select all" goes in beside it, asked of the catalog for the same reason:
the window is a hundred cells over a library of thousands, and a select-all
that quietly meant "the hundred that are loaded" is a lie the user cannot
see until the export runs. It sets the anchor to the first frame, or a
"Select to…" straight afterwards would reach `apply_press` with no anchor,
take its shift branch, and clear everything it had just taken.
The strip is at four buttons and a count now, so "New collection from
selection" loses its tail — the sheet it opens already says "New collection
holding 12 photographs" across the top.
Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
DarkRoom
A cross-platform, non-destructive RAW photo editor for Linux and Android.
Status: early. v0.1 is a remote library viewer — see docs/milestone-v0.1.md.
Documentation
| Document | Contents |
|---|---|
| requirements.md | What the software must do — 122 numbered requirements |
| architecture.md | How it is built — crates, GPU pipeline, data model, sync |
| milestone-v0.1.md | The first buildable milestone |
| faces.md | Face detection and identity — the models, the licence problem, and what S14 measures |
Building
Desktop:
cargo run -p darkroom-desktop
Android (containerised toolchain, see docker/android):
./docker/android/build.sh cargo ndk -t arm64-v8a build --release
Current state
Working: workspace, GPU context and compute pass, adaptive Slint shell, Android cross-compilation of the core crates.
Not yet working: the zero-copy display path. The build currently uploads frames through the CPU, which is exactly what ARCH §6.1 forbids — measured at 96% of frame time at 4K. Replacing it is spike S1, the project's highest priority.
cargo run -p dr-gpu --example bench --features readback
reproduces that measurement.
Licence
GPL-3.0-or-later.
Releases
20
DarkRoom 0.24.0
Latest
Languages
Rust
86.1%
Slint
10.3%
Python
1.1%
Shell
1%
WGSL
0.9%
Other
0.6%