Tom Lane wrote:
> Robert Haas <robertmhaas(at)gmail(dot)com> writes:
>> On Fri, Jan 29, 2010 at 3:32 PM, Heikki Linnakangas
>>> That only affects the error message and is harmless otherwise, but I
>>> thought I'd mention it. I'll fix it, unless someone wants to argue that
>>> its more useful to print the raw return value of system(), because it
>>> might contain more information like which signal killed the process,
>>> that you could extract from the cryptic error code using e.g WTERMSIG()
>> An "if" statement would seem to be in order, so that you can print out
>> either the exit code or the signal number as appropriate.
> Yes. Please see the existing code in the postmaster that prints
> subprocess exit codes, and duplicate it (or perhaps refactor so you can
> avoid code duplication; though translatability of messages might limit
> what you can do there).
Here's what I came up with. Translatability indeed makes it pretty hard,
I ended up copy-pasting.
BTW, I don't think I'm going to bother or risk back-patching this. It
was harmless, and for forensic purposes all the information was there in
the old message too, just in a cryptic format. And the new messages
would need translating too.
In response to
pgsql-bugs by date
|Next:||From: Jerry Gamache||Date: 2010-02-09 21:55:32|
|Subject: BUG #5321: Parallel restore temporarily deadlocked by autovacuum analyze|
|Previous:||From: Russell Smith||Date: 2010-02-09 07:31:11|
|Subject: Re: BUG #5319: recursion in the triggers|