On 2/27/22 15:23, Eric Pouech wrote:

Le 25/02/2022 à 20:19, Jacek Caban a écrit :

On 2/23/22 18:05, Eric Pouech wrote:

diff --git a/programs/conhost/conhost.h b/programs/conhost/conhost.h
index e5f70aa3ca4..35876689419 100644
--- a/programs/conhost/conhost.h
+++ b/programs/conhost/conhost.h
@@ -88,6 +88,7 @@ struct console
      unsigned int           read_ioctl;          /* current read ioctl */
      size_t                 pending_read;        /* size of pending read buffer */
      struct edit_line       edit_line;           /* edit line context */
+    unsigned int           key_state;
      struct console_window *window;
      WCHAR                 *title;               /* console title */
      struct history_line  **history;             /* lines history */

I think it would fit better in edit_line struct.

as the edit struct is zeroed upon completion (and before returning the result), it would also require moving

- write of the key state inside the edit buffer in process_console_input() (insert 2 new characters for the DWORD)

- hence handle in process_console_input that the layout of the buffer differs depending on ioctl

I thought it was more readable to keep the edit logic as is, and and on top of it returning or not the keystate flag

but see no reason why it wouldn't be doable

Agreed, it's probably not worth the complication. That's fine with me.


Thanks,

Jacek

ok I'll resend with the fix on the delta_y inversion (likely tomorrow depending on what Alexandre commits or not tonight)

A+