On Wed, Sep 9, 2009 at 6:04 PM, Tom Lane<tgl@sss.pgh.pa.us> wrote:
> Keh-Cheng Chu <kehcheng@solar2.Stanford.EDU> writes:
>> While 17 decimal digits are sufficient to guarantee the
>> complete recovery of all 64-bit double values, some 32-bit
>> reals actually require NINE significant figures.
>
> Hm, annoying. =A0Seems like we could do one of two things:
>
> 1. Increase the max extra_float_digits value to 3 and have pg_dump use
> that.
>
> 2. Split extra_float_digits into two settings so that float4 and float8
> can use different settings.
>
> Offhand the only argument I can see for #2 is that #1 might bloat dump
> files unnecessarily --- but it's not likely to be significant on a
> percentage basis.
>
> We'd also need to check the code to be sure it's leaving room for the
> extra digit in all cases, though I doubt there's really any problem.
(2) seems like overkill to me, FWIW.
...Robert
