Re: pgsql: Add 'basebackup_to_shell' contrib module.

On 3/29/22 17:19, Robert Haas wrote:
> On Tue, Mar 29, 2022 at 4:36 PM Thomas Munro <thomas(dot)munro(at)gmail(dot)com> wrote:
>> It failed:
>>
>> https://cirrus-ci.com/task/5567070686412800
>> https://api.cirrus-ci.com/v1/artifact/task/5567070686412800/log/contrib/basebackup_to_shell/tmp_check/log/001_basic_primary.log
>> https://api.cirrus-ci.com/v1/artifact/task/5567070686412800/log/contrib/basebackup_to_shell/tmp_check/log/regress_log_001_basic
> Hmm. The log here is not very informative. It just says that the first
> time we tried to use the 'shell' target, it timed out. I suppose the
> most obvious explanation for that is that the shell command we
> executed timed out:
>
> qq{type con > "$escaped_backup_path\\\\%f"}
>
> But why should that be so? Does 'type con' not respond to EOF? I don't
> see how that can be the case. Is our implementation of pclose broken?
> If so, then I think COPY TO/FROM PROGRAM would be broken on Windows.
>

AIUI 'type con' is not the equivalent of Unix cat, especially w.r.t.
stdin. It's going to try to read from the console, not from stdin. It's
more the equivalent of 'cat /dev/tty'. So it's not at all surprising
that it hangs. I don't know of a Windows builtin that is equivalent to cat.

cheers

andrew

--
Andrew Dunstan
EDB: https://www.enterprisedb.com