refactor: use the English json as the schema #41146
Conversation
|
|
|
When I added this validation script, the schemas were already being used to test (through a different approach) and had to be re-written to work with this approach. I vaguely remember thinking about using the English files as the single source of truth, but thought it was a bad approach. Can't for the life of me remember why I thought that, though. Regardless, the schemas are proving to be a bit of technical overhead that I agree we don't need (I forget to update them myself all the time). 100% in favour of removing them and using the English file as the root for testing. When this lands, I'll update the |
|
Yeah, I remember us discussing it briefly, but I can't remember what we concluded either. If this proves troublesome somehow we can always revert.
You can pop that onto this PR if you reckon it fits. |
|
I agree, Nick you should update this PR to update the validation script so it stays together. |
|
Just took a look at this - the However, I'm going to push an update to validate the other |
Validates the keys in all of the translation JSON objects. Note that this will return some false positives for dynamic keys (like the superblock intros). This is okay as the script is not meant to be used as blocking CI, but only to help us clean up keys as we update our client/API.
@nhcarrigan I was thinking about #41140 and I reckon this problem arose because it's not obvious to a human what
donate.confirm-3is and when it should be used instead ofdonate.confirm-2. That inspired me to rewrite them (confirm-monthly,confirm-one-timeetc.) and then I realised I'd need to rewrite the schema too.That leads me to this PR. Since we're trying to validate that all three languages have the same json structure, is it sufficient to validate that everything has the same structure as
english?