standardize translation method

[huss]

Is your feature request related to a problem? Please describe.

In many places OED uses translate() and in some it uses FormatMessage(). They both appear to return a string.

Describe the solution you'd like

OED should pick the better of these two and use it consistently in the code. Maybe research the two and propose one to the project. Then make the needed edits once everyone agrees.

Describe alternatives you've considered

Leave as is

Additional context

None

[huss]

This seems to relate to issue #1189 and can probably be done together.

[lawsj]

Please assign this issue to me.

[lawsj]

I found 15 instances of formatMessage() across the files. I found 66 instances of translate() across the files.

General uses:

Both in react-intl library

translate(): Convert strings from one language to another, using a file/directory as a key. ex) translate(“greeting”) returns “hello” for english or “hola” for Spanish

formatMessage(): Formatting strings from variables Good for displaying messages that use variables, such as:error messages, user notifications, etc. https://formatjs.io/docs/react-intl/api/ ex) FormatMessage({id: ‘welcome’, defaultMessage: ‘Welcome, {name}!’}, {name: ‘Alice’}) returns “Welcome, Alice!”

https://react.i18next.com/latest/usetranslation-hook

Experimenting: The changes from translate() to useTranslate() did not show errors, but all of the changes from formatMessage() to useTranslate() showed errors. These were the errors encountered when trying to implement this change:

1. Only allow translate() instead of useTranslate()
2. Intl would have to be removed from each instance, since the library is not used to call the useTranslate() function.

I think that it may be better to use the formatMessage() across the program if we want to continue to use the react-intl library. Translate() seems simpler, but it does not seem to be able to handle as many variables like the formatMessage() can. I would even propose to use translate() in some areas and use FormatMessage() strictly for error handling, which is how the instances seem to be used currently.

[nneni]

I like how thorough your analysis of the two functions was, it was very enlightening and helped me form a better, more factual opinion of the efficiency of the use of either function. I agree with your statement at the end as well, the functions seem to each work better in different situations.

[huss]

@lawsj Thanks for looking into this and your analysis. Would it be possible to meet to discuss your two options? (You can DM/email me to find a time.) I'd like to understand the tradeoff a little better so a final decision can be made. If not, I can ask questions here.