Anecdotal evidence is that most people are reading the csv files
line-by-line with TextIO and then parsing into columns in a subsequent
DoFn, ignoring (or asserting) that quoted newlines won't occur in their
data.
On Mon, Jun 18, 2018 at 11:27 AM Austin Bennett
wrote:
> Hi Beam Users/Dev,
>
> Ho
Hi Beam Users/Dev,
How are people handling currently handling CSVs as input to Beam (or not
really doing so)?
I see the things listed at the start of this thread -- any others?
I have many batch workflows involve getting multi-GB CSV files from third
party data aggregators (ex: hourly) and inges