Closed benoitjchevalier closed 3 months ago
Yep - you are right. We should add the first 3. I'd like advice on the braille ones from @cookiecrook @sinabahram and @pkra My first thought is that yes they should but we should add some text that states that an automated translation of these should not be attempted - but I would defer to any of the above on this.
Also, @joanmarie @michael-n-cooper should there be a new translatableString
(strawman name) value type in addition to the existing string
value type?
If so, then the translatable section could be generated from ReSpec.
Regarding the braille attr, I think non-braille values for aria-braillelabel
are probably okay to translate. I think it's unlikely that a translation of braille cell-conserving abbreviations (e.g. "btn" for "button" or "g" for "group") would be useful.
As for the more rarely used Braille Pattern values for those attributes, I think it's unlikely any existing translation engine would try to translate them. For example, even if you knew the language (English) aria-braillelabel="⠏"
could mean "p" in Grade 1 Uncontracted, or "people" in Grade 2 Contracted. It'd be impossible to back translate without knowing the table, so I don't think there is any risk of it in the near term.
So the choices regarding the Unicode Braille Pattern range are:
I lean toward option 2. There's very little need to mention it in the spec because it's unlikely to be a problem. Explaining the problem space just adds more complexity to the spec without much gain.
Looks like the translatable section is missing a few properties. It defines:
I think it's missing:
The braille ones are a bit more complex but my understanding is that they should also be translated since they won't be translated by assistive technologies:
I can do a PR if those additions make sense.