Hi, it seems the S3 source connector underwent a significant regression around v4, as there is no way to handle newline in text values and it breaks schema refreshes without an explicit error. Is there any known workaround? Or am I missing something? @Support Team

Hey Swan, at first glance, this does seem like a regression in v4. The v3 connector had a

newlines_in_values

option that handled this. When v4 was rewritten to use the file-based CDK, that option was silently dropped. The v4 CSV parser does handle newlines, but only when values are properly quoted per RFC 4180 (e.g.,

"value with\nnewline"

). Unquoted newlines break both reads and schema discovery. If you can control the upstream data, wrap any values containing newlines in double quotes. The v4 parser should handle those fine with no special config. If you can't control the quoting, there's no config-level fix in v4 right now. Hope that helps!

Thanks, I'll ask the upstream team if that's an option. If I may, it would be useful to have this written somewhere in the doc

I'll ping our docs team to see if we can add this in. Thanks Swan!

Could you link the github issue so I can 👍 and follow it please?

Thanks!