On Fri, Jan 2, 2015 at 06:56:47PM +0800, Craig Ringer wrote:
> On 12/31/2014 02:28 PM, Tom Lane wrote:
> > It might happen to work, if you're talking the same Intel-ish CPU type
> > on both ends ... but if it fails, don't call us.
>
> It might for 32-bit, but I'm pretty sure it's not going to for 64-bit
> because Windows is LLP64 and Linux is LP64, so sizeof(long) differs. See
> http://en.wikipedia.org/wiki/64-bit_computing .
>
> (Yet another area where Windows is annoyingly and pointlessly different,
> like use of UTF16 instead of UTF-8, though at least there they have the
> excuse of doing it first.)
>
> Just use a Linux box or a virtual machine to start a copy of the
> database, then pg_dump -Fc it and pg_restore to a new instance on the
> Windows box.
I would also worry about collation differences between Linux and
Windows. We have seen index problems with OS upgrades causing
mismatches, so I assume Windows and Linux might have even larger
mismatches. We have discussed a way to test for mismatches but have
no solution yet.
--
Bruce Momjian <bruce@momjian.us> http://momjian.us
EnterpriseDB http://enterprisedb.com
+ Everyone has their own god. +
