treewide: correct stale comments

- MuxRecorder claimed minSdk 35; it is 31.
- Adb claimed API 34 is the current Android max, next to
  compileSdk 35.
- Server claimed the openAbstract budget is wall-clock bounded below
  the e2e deadline; the deadline is per stream, so the worst case is
  not. Describe what actually holds instead.
- Pixhash contradicted itself about java.desktop availability.

4 files changed, 11 insertions, 12 deletions

diff --git a/app/src/main/java/invalid/lena/scrcpy/Adb.java b/app/src/main/java/invalid/lena/scrcpy/Adb.java
index 2909156..8e04327 100644
--- a/app/src/main/java/invalid/lena/scrcpy/Adb.java
+++ b/app/src/main/java/invalid/lena/scrcpy/Adb.java
@@ -59,9 +59,9 @@ public final class Adb extends AbsAdbConnectionManager {
 
     // libadb-android's mApi is documented as "the target's API for protocol
     // negotiation". We don't know the target's API at construction time;
-    // pinning to the current Android max means the protocol stays at its
-    // latest version, which is what wireless-debugging Android-11+ targets
-    // expect. Bump when newer protocol versions ship.
+    // pinning a recent release (API 34) keeps the protocol at the newest
+    // version wireless-debugging Android-11+ targets expect. Bump when a
+    // newer protocol version ships.
     private static final int  TARGET_API_HINT = android.os.Build.VERSION_CODES.UPSIDE_DOWN_CAKE;
 
     // Singleton: Main and Mirror both want an Adb; building two would
diff --git a/app/src/main/java/invalid/lena/scrcpy/MuxRecorder.java b/app/src/main/java/invalid/lena/scrcpy/MuxRecorder.java
index d06551e..b8d4aaf 100644
--- a/app/src/main/java/invalid/lena/scrcpy/MuxRecorder.java
+++ b/app/src/main/java/invalid/lena/scrcpy/MuxRecorder.java
@@ -11,7 +11,7 @@ import java.nio.ByteBuffer;
 // MP4 muxer that taps the H.264/H.265/AV1 elementary stream coming
 // straight from the scrcpy server - no re-encode, the bytes go onto
 // disk verbatim. AV1 muxing into MP4 needs Android 11+ (MediaMuxer
-// learned the codec there); we are minSdk 35 so that's fine.
+// learned the codec there); we are minSdk 31 so that's fine.
 //
 // State machine: IDLE → (arm) → ARMED → (next keyframe) → RECORDING
 //                                                          → (stop) → IDLE
diff --git a/app/src/main/java/invalid/lena/scrcpy/Server.java b/app/src/main/java/invalid/lena/scrcpy/Server.java
index cc16b44..6ae752f 100644
--- a/app/src/main/java/invalid/lena/scrcpy/Server.java
+++ b/app/src/main/java/invalid/lena/scrcpy/Server.java
@@ -40,11 +40,10 @@ public final class Server {
     private static final int    FILE_MODE      = 0100644;         // regular file, 0644
     // Server forks a CleanUp helper before opening its abstract sockets;
     // on slow emulators that takes several seconds, so the budget needs
-    // to be generous. Wall-clock bounded: even if every attempt wedges
-    // for the full OPEN_ATTEMPT_TIMEOUT_MS, the total stays near
-    // OPEN_DEADLINE_MS - strictly below the e2e test deadline
-    // (test-rig/e2e.sh's E2E_DEADLINE, default 60 s), or the test
-    // would fail the wrong way.
+    // to be generous. The deadline is per stream: in practice only the
+    // first open waits (the server is still starting) and the other two
+    // dial instantly, so a healthy bring-up stays well below the e2e
+    // test deadline (test-rig/e2e.sh's E2E_DEADLINE, default 60 s).
     private static final long   OPEN_DEADLINE_MS     = 20_000;
     private static final int    OPEN_BACKOFF_MS      = 100;
     // Per-attempt timeout for adb.openAbstract. libadb-android's
diff --git a/test-rig/Pixhash.java b/test-rig/Pixhash.java
index 47ab8b8..b35a560 100644
--- a/test-rig/Pixhash.java
+++ b/test-rig/Pixhash.java
@@ -5,9 +5,9 @@
 //   java test-rig/Pixhash.java <png>            # default min=5%
 //   java test-rig/Pixhash.java <png> --min 1
 //
-// We avoid AWT's BufferedImage because some headless JDKs ship without
-// it; ImageIO is in java.desktop which IS present in the openjdk-17-jdk
-// package in our image, so it's fine.
+// ImageIO/BufferedImage live in java.desktop, which the image's
+// openjdk-17-jdk-headless package does ship (headless only disables
+// display/input, not imaging) - no extra dependency needed.
 
 import java.io.File;
 import java.io.IOException;