Wallper 1.0.1: Two Days After Launch, the First Fixes
The first patch after release: a hardened apply pipeline, preview grids that cope with slow connections, and better diagnostics for first-run problems.
Two days after 1.0 went out, we learned the thing every developer learns roughly two days after shipping.
Your test machines are not representative. They are fast, they are on good connections, they have run every build of the app in order, and they are configured by the person who wrote it. Real Macs are older, slower, connected to hotel wifi, and encountering the app for the first time in a state your machine has not been in since the first week of development.
1.0.1 is the patch that fixes what that first weekend surfaced.
The Apply Pipeline
Applying a wallpaper looks like one action and is really several: resolve the file, hand it to the rendering layer, replace what is currently on screen, and persist the choice so it survives a restart.
Those steps have an order, and on a slow machine the gaps between them get wide enough for things to arrive out of sequence. The result was a rare failure where the wallpaper did not take, or took and then reverted, with no error to explain it.
The pipeline is now hardened against those timing issues. The steps hold their order regardless of how long each one takes.
Preview Grids on Slow Connections
The library grid assumed previews would arrive promptly. On a fast connection they do.
On a slow one, that assumption produced a grid that looked broken: empty cards, half-loaded thumbnails, and a layout that shifted under the cursor as images turned up in whatever order the network delivered them. Nothing was actually failing, but it looked like everything was.
The grid now degrades properly. Slow previews arrive late without disturbing the layout, and a card that is still loading looks like a card that is still loading rather than a card that has given up.
Crash Guards and Better Logs
Two defensive changes.
Missing metadata is now handled rather than assumed away. A wallpaper with an incomplete record used to be able to take down the view it appeared in; now it is skipped and the rest of the grid renders.
Logging around first-run behaviour is substantially better. Most of what people reported that weekend happened exactly once, on a fresh install, in a state that is difficult to reproduce on a machine that already has the app configured. Better diagnostics is how those get fixed rather than guessed at.
If you hit something, the troubleshooting guide is the place to start, and reports are what turn a rare bug into a fixed one.
