search

keymanapp / keyman

Keyman cross platform input methods system running on Android, iOS, Linux, macOS, Windows and mobile and desktop web
https://keyman.com/
Other
377 stars 104 forks source link

feat(windows): Investigate if VKContextReset is needed in for core on kmx processor #10227

Closed rc-swag closed 2 months ago

rc-swag commented 6 months ago

In reviewing #10090 it become apparent that the call to reset the context in the windows engine base upon the table VKContextReset was no longer used. This ticket is to investigate if it is still needed if so integrated into the core. Also remove the Array from the Windows engine.

rc-swag commented 6 months ago

I have tested some of the keys by running keyman. However, it is hard to test when the context would be reset anyway (focus changing). The pause key seemed to follow and the escape key but others I couldn't be sure. It might be best to have a core level unit test.

rc-swag commented 3 months ago

I have run the test host program and tested the keys available to me on my keyboard. There are a number of keys that reset the context that had a value of '0' in the VKContextReset array. See --- Reset Context when not expected by VKContextReset. I tested using the Euro Latin keyboard therefore using the kmx keyboard processor. The key to be determined whether we reset the context for the KMX processor seems to be line 262 of kmx_processevent.cpp if(!gp || (m_state.charCode == 0 && gp->fUsingKeys)) plus it not being a backspace. This essentially means it is reset if the key is not a charCode according to this struct char_to_vkey

Note: This table is replicated in the core vkey_to_contextreset kmx_processevent.h It is never called now even by the ldml_processor. 

1,   //L"K_BKSP",              // &H8
1,   //L"K_TAB",               // &H9

1,   //L"K_ENTER",             // &HD

0,   //L"K_SHIFT",           // &H10
0,   //L"K_CONTROL",    // &H11
0,   //L"K_ALT",                // &H12
1,   //L"K_PAUSE",              // &H13
0,   //L"K_CAPS",               // &H14

1,   //L"K_ESC",                // &H1B
0,   //L"K_SPACE",              // &H20
1,   //L"K_PGUP",               // &H21
1,   //L"K_PGDN",               // &H22
1,   //L"K_END",                // &H23
1,   //L"K_HOME",               // &H24
1,   //L"K_LEFT",               // &H25
1,   //L"K_UP",                 // &H26
1,   //L"K_RIGHT",              // &H27
1,   //L"K_DOWN",               // &H28

0,   //L"K_INS",                // &H2D                     ----- RESET Context
1,   //L"K_DEL",                // &H2E
0,   //L"K_0",                  // &H30
0,   //L"K_1",                  // &H31
0,   //L"K_2",                  // &H32
0,   //L"K_3",                  // &H33
0,   //L"K_4",                  // &H34
0,   //L"K_5",                  // &H35
0,   //L"K_6",                  // &H36
0,   //L"K_7",                  // &H37
0,   //L"K_8",                  // &H38
0,   //L"K_9",                  // &H39

0,   //L"K_NPSTAR",             // &H6A            ----- RESET Context
0,   //L"K_NPPLUS",             // &H6B             ----- RESET Context
0,   //L"K_SEPARATOR",          // &H6C         -------couldn't test
0,   //L"K_NPMINUS",            // &H6D           ----- RESET Context
0,   //L"K_NPDOT",              // &H6E           ----- RESET Context
0,   //L"K_NPSLASH",            // &H6F           ----- RESET Context
1,   //L"K_F1",                 // &H70
1,   //L"K_F2",                 // &H71
1,   //L"K_F3",                 // &H72
1,   //L"K_F4",                 // &H73
1,   //L"K_F5",                 // &H74
1,   //L"K_F6",                 // &H75
1,   //L"K_F7",                 // &H76
1,   //L"K_F8",                 // &H77
1,   //L"K_F9",                 // &H78
1,   //L"K_F10",                // &H79
1,   //L"K_F11",                // &H7A
1,   //L"K_F12",                // &H7B

0,   //L"K_NUMLOCK",            // &H90        ----- Didn't even go to engine
0,   //L"K_SCROLL",             // &H91         ----- RESET Context

0,   //L"K_COLON",              // &HBA
0,   //L"K_EQUAL",              // &HBB
0,   //L"K_COMMA",              // &HBC
0,   //L"K_HYPHEN",             // &HBD
0,   //L"K_PERIOD",             // &HBE
0,   //L"K_SLASH",              // &HBF
0,   //L"K_BKQUOTE",            // &HC0
rc-swag commented 3 months ago

I ended up adding logging to test the LDML processor. For non-compliant apps (non-context aware), it is never resetting context for a key press from the list above (well for any key). I realise now this has occurred because the action struct that the LDML processor is giving to the core does not have the option to invalidate the context.

Both LDML and KMX never inserted the \ - * keys from the number pad into the context, I am pretty sure they both are hitting the emit keystroke code. However, KMX resets context where as LDML does not.

We have 2 issues.

  1. KMX does resets context if the key is not a char code according to char_to_vkey this maybe acceptable the differences are the numlock * - \ keys and the scroll lock, and ins key.
  2. The action struct from processor to core does not provide a mechanism for invalidating/resetting the context.