Re: write_pipe_chunks patch messes up early error message output

Tom Lane wrote:
> Andrew Dunstan <andrew(at)dunslane(dot)net> writes:
>
>> Tom Lane wrote:
>>
>>> I think offhand that the correct semantics of the flag are "we have
>>> redirected our original stderr into a pipe for syslogger",
>>>
>
>
>> We could expose syslogger's redirection_done flag, which I think has the
>> semantics you want.
>>
>
> Yeah, that would work. You'd have to get rid of the current ad-hoc
> method by which it is propagated to the syslogger child process
> (EXEC_BACKEND case), because now it will have to be propagated to all
> children; so postmaster.c should handle it in BackendParameters.
>
>
>

The problem with this as it stands is that the syslogger itself is
forked before the redirection is done. I guess we need to make sure the
syslogger itself never calls write_pipe_chunks() - which makes sense
anyway - should probably call write_syslogger_file() directly, I think.

cheers

andrew