Inconsistent behaviour with carrot

[seekM]

Carrot:

 "carrot":
    R(Text("^"), rdescript="Core: Carat"),

I'm getting inconsistent behaviour using carrot. I was just trying it and saying carrot yields nothing, saying it again yields ^^. Yesterday saying carrot gave me ^^ directly on the first try. Expected behaviour: Say it once, get ^. I tried it in Notepad++. Anyone else experiencing this?

[kendonB]

No unusual behavior here. Are you using an unusual keyboard? On my Māori keyboard, we prefix with the backtick to add macrons, for example, which introduces unusual behavior if you're actually trying to make a backtick.

[kendonB]

Is there another carrot command somewhere deep in your version of caster? try find in files?

[seekM]

I fiddled with Window 10 language and keyboard settings and figured this out:

carrot output depending on...
German language, German keyboard: ^^
German language, English US keyboard: backtick (can't format it here)
English language, German keyboard: &
English language, English US keyboard: ^

[seekM]

For Windows 10 German language, German keyboard layout I could fix it using this rule filter:

def update_carrot(rule):
    if "carrot" in rule.mapping_actual().keys():
        rule.mapping_actual()["carrot"] = R(Key("caret") + Key("left") + Key("delete"))

def fix_carrot_german_windows10_scenario(mp):
    if mp.time == MergeInf.BOOT:
        if mp.rule1 is not None :
            update_carrot(mp.rule1)   
        update_carrot(mp.rule2)

So this issue can be closed I think, but it should probably be added to the documentation that you have to look out for these kind of issues and might need modifications depending on which language / keyboard layout you have.

[kendonB]

Can we detect the keyboard and language within caster and fix up the output? Long term it would be a shame if we only supported the English / English combo.

[LexiconCode]

Yes we do need to handle alternate keyboards better. Fundamentally though is this something that's a caster issue or dragonfly?

[kendonB]

A KeyboardLanguageContext class from dragonfly would be great. People do change these on the fly (at least I do) so we'd want to detect last second like AppContext

[kendonB]

Will close this for now while we wait and see what comes from Dragonfly