On 22 March 2015 at 14:46, Bruce Momjian <bruce@momjian.us> wrote:
On Wed, Mar 18, 2015 at 05:52:44PM -0400, Bruce Momjian wrote: > In September, while researching the to_char() buffer overflow bugs fixed > in 9.4.1 (commit 0150ab567bcf5e5913e2b62a1678f84cc272441f), I found an > inconsistency in how to_char() does zero-padding for float4/8 values. > Now that 9.4.1 is released and I am home for a while, I am ready to > address this. ... > float4/8 are padding to the internal precision, while int4/numeric are > padding based on the requested precision. This is inconsistent. > > The first attached patch fixes this, and also zeros the "junk" digits > which exceed the precision of the underlying type:
Patch applied.
This seems to have broken jacana. Looks like MSVC by default has a 3 digit exponent.