Update match pattern and translation rules for "curl" in calculus.yaml#432
Conversation
|
Answering questions in order:
That's being defensive, but it very likely won't matter in practice. MathCAT will only ever generate "curl" (and "div", etc) with one child. However, they can be created with MathML 4 via I've been moving toward a simpler system for "simple" rules like this. They are defined as one-liners in
I've avoided using capital letters for fear the speech engines would distinguish their speech in some way. NVDA and JAWS both have options for how capital letters should be spoken while navigating. But your change made me go back and try using them for when speaking words, not letters during navigation and at least in my tests with NVDA, they don't get spoken any differently. So it is OK to use capitals. Proper capitalization won't affect the speech and so isn't noticeable in that sense, but if you copy/paste the speech (an option in NVDA and maybe JAWS), I can see how it would be better for it to be proper German. It took a few tries before I was convinced that the speech wasn't affected, hence the delay in approving this.
Yes, as long as it makes sense. There are some languages where the translators never added "the" or never added "of". There may be cases where in English, there isn't a verbose form for something but in other languages there is. What gets said, and what happens for each option, is something that I leave to the translator to decide.
I added phrases in the hope that having some context would improve google's translation. It hasn't helped anywhere near as much as I had hoped. But maybe it will help for DeepL? The way it works is that instead of sending like 'curl' to google and getting back a translation, the script sends "the 'curl of' a field" and then looks at the result and picks out the translated word inside the '...'. Sadly, sometime what comes back doesn't have single quotes. That is one of the things that has made automating the translations harder. |
This is my first minor PR to get started with the rules for German.
My thoughts:
en/SharedRules/calculus.yamlhave the conditionmatch: "count(*) = 1", while the one for German so far uses"."? I think it makes sense to only match when there is exactly one child, in both languages."$Verbosity!='Terse'", while the German one does not. Generally, this should be supported as much as possible, right?# phraseshould always be an example for the language in use, so is this just a copy-paste error?