ibus-typing-booster always uses NFC for the display of the preedit and always commits everything converted to NFC.
so when typing e _ H space, then a é U+00E9 LATIN SMALL LETTER E WITH ACUTE is commited.
When the commit happend, the cursor displayed in the preedit was at the end of the string é, i.e. at position 1 in a string with length 1, i.e. at the right end of the preedit. So no cursor position correction should be done between commiting the é and committing or forwarding the space.
But the current cursor position correction mistakenly uses the NFD version of the commit string, which has length 2. So it tries to "correct" the cursor position by moving one step left after commiting the é which puts the cursor before the é and then the space is inserted before the é .
mike-fabian
commented
5 months ago
This input method https://github.com/mike-fabian/m17n-db-ipa-x-sampa/blob/main/ipa-x-sampa.mim#L114 sometimes produces NFD output, for example when typing
e_Hthe output ise U+0065 LATIN SMALL LETTER Efollowed byU+0301 COMBINING ACUTE ACCENT.ibus-typing-booster always uses NFC for the display of the preedit and always commits everything converted to NFC.
so when typing
e _ H space, then aé U+00E9 LATIN SMALL LETTER E WITH ACUTEis commited.When the commit happend, the cursor displayed in the preedit was at the end of the string
é, i.e. at position 1 in a string with length 1, i.e. at the right end of the preedit. So no cursor position correction should be done between commiting theéand committing or forwarding the space.But the current cursor position correction mistakenly uses the NFD version of the commit string, which has length 2. So it tries to "correct" the cursor position by moving one step left after commiting the
éwhich puts the cursor before theéand then the space is inserted before theé.