[packetThrower]

Fixes from testing v0.8.0-beta.1 on Windows with an MX25L12835F (SPI) and a 24C64 (I²C).

SPI / Macronix

• Status registers: Macronix chips are decoded with their own layout (BP0–3, QE, SRWD), and the configuration register replaces SR3. SR1 0x40 (QE) no longer warns about block protection. 0x35 is no longer sent to Macronix parts, where it is EQIO and switches the chip into QPI mode.
• Security registers: the ops refuse Macronix, which uses a different OTP scheme, and the pane hides its controls.

Errors and files

• A failed or cancelled read deletes the partial dump. The file is closed first so Windows can delete it.
• A CH341A that is plugged in but can't be opened reports the driver / udev fix instead of "not found".
• Unplugging mid-operation reports "disconnected" instead of "Pipe error".

I²C

• Changing the chip updates the header, disarms a pending write or erase, and clears the stale result message.
• Arming a write refuses a file larger than the chip, and the armed log line names the file, its size and the chip.

UI

• The View menu has Zoom In / Out / Actual Size, and the menu shows "Ctrl+=".
• File dialogs open in the read save location.
• The sidebar scrolls when the window is short.
• Log lines wrap before the toolbar icons.
• The hex description is shorter.
• The BIOS explorer explains non-UEFI images.

Testing on Windows

• MX25L12835F reads match the known hash, including after a mid-read unplug.
• The status decode is correct, and Detect still works afterwards, so the chip didn't enter QPI.
• Hex zoom works from both the keys and the menu.
• 10/10 I²C write/verify cycles were clean.

One intermittent I²C first-page drop was seen once on an earlier build and didn't reproduce in 10 cycles; it's tracked separately.

🤖 Generated with Claude Code