Tom Lane wrote:
> Andrew Dunstan <andrew(at)dunslane(dot)net> writes:
>> My thought is to have the logger mark each true log line line end with a
>> null byte.
> This seems not to help much when you consider interleaved output from
> different backends...
True. We have two problems with log robustness, and this was an attempt
to address one of them. I sent some thoughts about interleaving
yesterday. If you have any thoughts about handling interleaving then I'm
all ears. However, my impression was that this only happened on a small
number of platforms. If it's more general then it's all the more urgent
to fix. ISTM that we can't really consider CSV logging until we have a
handle on both problems.
In response to
pgsql-hackers by date
|Next:||From: Joe Conway||Date: 2007-06-06 03:02:51|
|Subject: Re: Implicit casts with generic arrays|
|Previous:||From: Tom Lane||Date: 2007-06-05 22:00:54|
|Subject: Re: more robust log rotation |