On 19 Jul 2014, at 4:42pm, James K. Lowden <[email protected]> wrote:
> Jonathan Leslie <[email protected]> wrote: > >> What I do is I never use a CSV file as a "Comma separated Values" >> file but rather as a "Character separated Values" file, but rather >> use a Character, ยบ (<alt>0186), a legal character but not one on the >> keyboard, as the separator character. > > Better might be to use ASCII as designed. It defines position 30 as a > record separator. I'm clarifying that only to stop someone writing wrong code, not to be annoying. ASCII 30 is Record Separator. Use this where you might expect a newline. ASCII 31 is Unit Separator. Use this where you might expect a comma. Files which used these codes (together with 28 and 29) used to be called 'Control Separated Data' files. So you used to see files with names like SENSOR6.CSD . In SIXBIT, of course. By the way, two US characters together indicates a blank field. But two RS characters together does not indicate a blank row. You carry on as if any number of RS together are all one RS. So you can delete a record without having to change the length of a file by replacing all the characters in the record with RS. This was useful when removing the middle of a file could occupy a disk or tape drive for 20 minutes. And yes, I'm old enough that I used to write programs that ran on mainframes that used these codes. Simon. _______________________________________________ sqlite-users mailing list [email protected] http://sqlite.org:8080/cgi-bin/mailman/listinfo/sqlite-users
