Re: Fixing backslash dot for COPY FROM...CSV

Bruce Momjian <bruce@momjian.us> writes:
> On Sun, Apr  7, 2024 at 12:00:25AM +0200, Daniel Verite wrote:
>> Agreed we don't want to document that, but also why doesn't \. in the
>> contents represent just a dot  (as opposed to being an error),
>> just like \a is a?

> I looked into this and started to realize that \. is the only copy
> backslash command where we define the behavior only alone at the
> beginning of a line, and not in other circumstances.  The \a example
> above suggests \. should be period in all other cases, as suggested, but
> I don't know the ramifications if that.

Here's the problem: if some client-side code thinks it's okay to
quote "." as "\.", what is likely to happen when it's presented
a data value consisting only of "."?  It could very easily fall
into the trap of producing an end-of-data marker.

If we get rid of the whole concept of end-of-data markers, then
it'd be totally reasonable to accept "\." as ".".  But as long
as we still have end-of-data markers, I think it's unwise to allow
"\." to appear as anything but an end-of-data marker.  It'd just
add camouflage to the booby trap.

            regards, tom lane