Michael Paquier <michael@paquier.xyz> writes:
> On Sat, Oct 26, 2019 at 11:02:26AM -0700, Adrian Klaver wrote:
>> Not sure how that can be answered without knowing what ComputeComputer is
>> doing?
> Yes, there is nothing of this kind in the PostgreSQL code.
Sure there is: win32error.c produces exactly that message if its
mapping table has no entry for the Windows error code.
According to
https://docs.microsoft.com/en-us/windows/win32/debug/system-error-codes--0-499-?redirectedfrom=MSDN
code 38 decimal is ERROR_HANDLE_EOF, "Reached the end of the file.",
which indeed is not in the doserrors[] table.
But that just deepens the mystery --- if we hit an EOF, why wasn't
that handled through more-normal channels? We still need to know
what that function was trying to do, and whether any non-core
C code was involved.
regards, tom lane