[schnierk]

What

dev_icon::tests::desktop_file_matches_the_packaged_fields fails on main.

69a6c4b ("packaging: Linux file managers offer FilmCraft for .fcproj projects and video files") added a MimeType= line to packaging/linux/ai.storyteller.filmcraft.desktop. The opt-in dev desktop entry that dev_icon::desktop_file_content writes still has the old field list. The test is there so the two cannot drift; it now compares them and fails.

Fix

The dev entry writes the same MimeType= line as the packaged one, in the same place. With FILMCRAFT_DEV_DESKTOP_ENTRY=1, a source build is then offered for projects and video files like the installed app, which is what the test's contract ("the entry mirrors the packaged one") says.

Tests

• cargo test --release -p filmcraft --bin filmcraft dev_icon: 6 passed (it failed before).
• cargo fmt --check passes.

Found while running cargo xtask ci for #1957 on Linux. Prepared with Claude Code (Opus 5.5).

🤖 Generated with Claude Code

[FredApps]

This is right, and it is the smallest way to fix it. packaging/linux/ai.storyteller.filmcraft.desktop on main has a MimeType= line, and dev_icon::desktop_file_content does not write it, so desktop_file_matches_the_packaged_fields fails: it compares the whole generated file with the packaged one line by line (dev_icon.rs:272-283), minus TryExec= and with Exec= replaced. I compared the added line with the packaged one and the 12 MIME types are the same, in the same order, and the line sits between Categories= and Keywords= as in the packaged file, so the exact comparison holds.

Hosted CI does not build or test filmcraft, so the failure only shows when someone runs the suite. I am leaving this as a comment, with a request for a receipt for this exact head.

This is also the PR the others should defer to. The same one-line change is carried by #2092, #2122, #2012, #1980 and #1978, and #2370 (a Linux xtask ci fix) also touches dev_icon.rs for the same failing test; #1958 is the one that does only this. Whichever lands first makes the others conflict, so the authors of those should drop the hunk. I will say so on them as I get to them.

One small point: the line is a second copy of a long MIME list, so the next change to the packaged file breaks the test again. The test is what keeps the two in step, so that is acceptable; I would not ask for more here.

Validation request (head e225371cae16e74522d43d0445c7cc02bf65e57e):

• os: linux
• commands: cargo fmt --all -- --check; cargo clippy --workspace --all-targets --locked --offline -- -D warnings; cargo test -p filmcraft --bin filmcraft --locked --offline dev_icon
• expected: filtered run 6 matched, 6 passed, 0 ignored (as the PR text says, desktop_file_matches_the_packaged_fields was the failing one)
• holds: none (clippy nonminimal_bool on main, #2007, is not caused by this PR)