Alfred-Mountfield commented on pull request #12451: URL: https://github.com/apache/arrow/pull/12451#issuecomment-1060572661
Apologies for the late response. Thank you for pushing for more concrete proof that this PR achieves its intention. While trying to create a better test-case for you I realise we have a large quantity of errors within our uses of the library that have only been exposed to me while trying to create a test-case within the arrow project itself in an IDE with proper intellisense. I think the code examples I have given (i.e. the StackBlitz links) aren't valid due to errors such as these: <img width="1119" alt="image" src="https://user-images.githubusercontent.com/25749103/157021673-7e14f2ae-ca97-4321-8fae-2bf64dcf8745.png"> <img width="537" alt="image" src="https://user-images.githubusercontent.com/25749103/157021284-5a0c8df4-c424-437a-87ff-944d80a22975.png"> <img width="699" alt="image" src="https://user-images.githubusercontent.com/25749103/157021782-25003004-84fc-4a97-b0a6-711ff7fbf0c9.png"> Unfortunately, using the JS Arrow packages don't really give any indication that doing things like that are incorrect. Not trying to restate the obvious benefits of using TypeScript, but I do worry a bit that the arrow package lends itself to so many errors that fail somewhat silently in plain JS because of things like abstract classes. I think having better documentation showing more use-cases and trying to convey convention would help a lot with this. Is there somewhere I can raise this properly (as this PR doesn't feel like the right place to start that discussion), should I just open a Jira ticket or will that likely languish? Going back to this PR, I'm going to investigate the methods a bit more deeply, and try and understand the library's implementation a bit better when it comes to things like `setValid` etc. I worry that the spec is a little too vague when it comes to the specifics of the lengths of buffers in things like `fixedSizeList`s and `Struct`s with null elements, and it's possible the different language impls have conflicting interpretations. I'll do my investigation in TS (rather than JS as I was before) and post again afterwards :) -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: github-unsubscr...@arrow.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org