Hacker News Reader

I think it's valid to argue that you shouldn't be able to put some of commas, quotes, and newlines inside fields at all. And comma versus semicolon.

But that doesn't extend to using backslash escapes in something that's legitimately trying to be CSV. That's someone getting confused and implementing a mix of data formats, or trying to be clever and making an extended CSV format.

View on HN

It’s valid to argue that, but that means you can’t use CSV for many real-world data sets.

That, in turn, means you almost cannot use CSV in any robust solution. Even if, today, your input doesn’t have commas, quotes or newlines, can you guarantee it won’t tomorrow, next year, etc?

Yes? What do I not know that makes this question harder than it seems to me?

> Even if, today, your input doesn’t have commas, quotes or newlines, can you guarantee it won’t tomorrow, next year, etc?

But... but those are the ones I listed as real special characters, unlike backslash. I don't understand the question.

Every field is quoted. Every field is quote-escaped. That's all you need.

There are simpler and better approaches. Fortunately for us all, IP datagrams are not JSON-encoded, for example.

Reply on news.ycombinator.com