David G. Johnston wrote: > I'm not following - if recordsep is not something that would > interpreted as a newline then the file we output would have not > structural newlines. > It might have data newlines but those would be quoted.
They would be, but I don't quite understand the point in your first sentence. All I'm saying is that setting recordsep to '\r\n' does not achieve the goal of obtaining CRLF line endings. Here's an example with the unaligned mode, linux host: (psql -At -P recordsep=$'\r\n' << EOF select E'A\nB' union E'C\nD'; EOF ) | hexdump -C The result is: 00000000 41 0a 42 0d 0a 43 0a 44 0a |A.B..C.D.| The expectation of CRLF line endings is that every LF would be preceded by CR, but here we get that only for 1 LF out of 4. That's useless. It's not a bug. We asked for a CRLF to separate our two records and that's exactly what we got, no more no less. In csv output, the difference would be that there would a double quote character before A and after B, and before C and after D but otherwise it would be the same mix of LF and CRLF. I think that the point of recordsep in unaligned mode is you can set it to something that never appears in the data, especially when embedded newlines might be in the data. In CSV this is solved differently so we don't need it. Best regards, -- Daniel Vérité PostgreSQL-powered mailer: http://www.manitou-mail.org Twitter: @DanielVerite
