"Andrew Dunstan" <andrew@dunslane.net> writes:
> I wish I could understand why people are so keen to make other people turn
> handsprings in order to avoid a feature which, as Bruce points out, is
> already on the TODO list, and which, by my 10 minute analysis, would involve
> almost trivial code impact and risk. If this involved major impact I might
> understand, but it really doesn't.
Supporting all of the CSV options in pg_dump would involve major bloat
in its option set, and it already has far too many options. If it were
just a matter of adding a "--csv" switch I wouldn't be complaining, but
there are half a dozen more sub-options, and it seems like every time we
turn around someone is finding a reason for another one. Propagating
all that cruft through pg_dump would be a PITA, and updating it to track
future additions would be too.
Furthermore, the entire rationale for the feature is predicated on the
claim that programs other than pg_restore might find it useful. But
this conveniently ignores the fact that if there are any such programs
in existence, what this will really do is BREAK them, because they won't
be able to cope with all the variants that pass for CSV.
My opinions would be less negative if I thought that CSV were a
well-defined format that would never change. I don't believe that it
has either property, however, and so I'm against letting it get into our
dump file format. I think we'll just live to regret it if we do.
regards, tom lane
