Windows support: not now, maybe never

Greenhouse’s PRD has quietly committed to macOS + Windows since early on, mostly because I liked the idea of one day handing Mike Monday’s students a friendly cross-platform build. But an earlier issue deferred the actual Windows CI job to “later,” and later never got filed as its own issue. So the commitment sat there, untracked, doing nothing.

I finally made myself pick: drop it from v1. Tauri still supports Windows architecturally, nothing in the current code choices forecloses it, so if a Windows build actually becomes worth building later (real demand, or the teaching-template idea turning into a real next step instead of a maybe) it’s still on the table. But right now it was unscoped work sitting in the PRD as an implicit promise, with zero path-handling audit, zero CI, and zero pressure to actually build it.

Updated the PRD to say macOS-only for v1, with a one-line pointer back to the decision. Fifteen minutes of work, mostly documentation, and one fewer piece of untracked scope hanging over the release pipeline.

The general shape is worth naming: a commitment that lives only in a planning doc, with no issue tracking it and no work scheduled against it, isn’t a plan. It’s a liability that makes every estimate quietly wrong. Deciding “not now, maybe never” out loud costs almost nothing and returns the scope to reality.