>>>>> "Andrew" == Andrew Dunstan <andrew.dunstan@2ndquadrant.com> writes:
>> [horrible environment variable hack]
>>
>> We might not want to use that in more real-world cases of pg_dump use,
>> but I think for this purpose it should be fine.
Andrew> I haven't seen a response to this. Cross version upgrade
Andrew> testing is still broken. I don't think we need a GUC to fix it,
Andrew> but we do need this or a new switch to tell pg_dump what to set
Andrew> extra_float_digits to, unless someone has a better idea.
I'd been holding off responding in the hope of other opinions, but for
what it's worth, I *really* dislike having pg_dump depend magically on
some new environment variable. I would suggest instead:
a) pg_dump could check if PGOPTIONS or the connect string contained an
extra_float_digits setting and defer to that if so.
Downside of this is that if someone is already using that in the
environment and pg_dump suddenly starts respecting it, they could get
imprecise values in their dumps unexpectedly. Option (a2) would be to
honour extra_float_digits only if it showed up in a connect string and
not in PGOPTIONS, which would be more explicit.
b) new command-line option, e.g. pg_dump --extra-float-digits=0
This is probably the safest option, IMO. Any preferences as to the
option name?
--
Andrew (irc:RhodiumToad)