On Tue, Jul 14, 2020 at 12:54:58PM +0900, Michael Paquier wrote:
> On Mon, Jul 13, 2020 at 08:06:25PM -0400, Bruce Momjian wrote:
> > Well, pg_test_fsync is testing binary writes, so using TEXT for the
> > files might add unwanted overhead, so I think binary is the best
> > approach.
>
> Thanks. Please note that we have switched frontend tools to use our
> concurrent-safe flavor of open() and fopen() as of 0ba06e0 (12 and
> newer versions), but 40cfe86 has shown that we have been using the
> text mode in pg_test_fsync since forever as Windows' open() uses the
> text mode by default if we don't specify _fmode with _setmode():
> https://docs.microsoft.com/en-us/cpp/c-runtime-library/fmode?view=vs-2019
>
> For this reason, it seems more sensible to me to not backpatch this
> change only down to 12, but actually all the way down to 9.5.
OK. I am kind of surprised we haven't received bug reports earlier, but
I guess few Windows users run pg_test_fsync, and fewer report problems.
--
Bruce Momjian <bruce@momjian.us> https://momjian.us
EnterpriseDB https://enterprisedb.com
The usefulness of a cup is in its emptiness, Bruce Lee