Peter Eisentraut <peter_e@gmx.net> writes:
> Tom Lane writes:
>> echo $version > "$PGDATA/PG_VERSION"
> No, remember that this is new for 7.1. 7.0 uses
Oh, right, you rejiggered all that stuff a few months back. My mistake.
> sprintf(version, "%s.%s\n", PG_RELEASE, PG_VERSION);
> fd = open(full_path, O_WRONLY | O_CREAT | O_EXCL | O_BINARY, 0666);
> write(fd, version, strlen(version));
> which should not cause newline foul-ups.
Actually, the 7.0 code seems to apply O_BINARY only when __CYGWIN32__
is defined; I wonder if the NT version somehow got built without that?
>> Of course this just begs the question of why it works for some people
>> on NT and not others. Is this a difference across cygwin releases,
>> perhaps?
> My understanding is that this is settable by mount-point, so the confusion
> is practically unsurmountable.
But surely O_BINARY would prevent a newline translation from occurring
no matter what?
Or is it possible that Alexander tried to copy the files from one disk
to another, and the copy operation decided to do newline translations?
I still don't understand why he's seeing a failure...
regards, tom lane