enchant2 configuration lacks many extra-word-chars
When doing spellcheck in gnome-text-editor (and presumably other applications using enchant2 directly or via gspell), some characters outside the language's natural breaking characters need to be included to determine proper word boundaries.
The way enchant2 is configured in org.gnome.Platform results in seeing from enchant_dict_get_extra_word_characters()
a value like "0123456789"
. Where as in Fedora you'll get something like "-0123456789abcdef..."
.
That means when using gnome-text-editor
from Flatpak, we'll get poor spellcheck results due to seeing:
"isn't" => ["isn", "t"]
when on Fedora we would see something like
"isn't" => ["isn't"]
For now, I just manually work around this in gnome-text-editor@6266ed2a but that's probably not a great idea for non-latin based languages.
Fedora's SPEC file for reference: https://src.fedoraproject.org/rpms/enchant2/blob/rawhide/f/enchant2.spec