Re: Re: [COMMITTERS] pgsql: Map basebackup tablespaces using a tablespace_map file

On Thu, Sep 10, 2015 at 3:52 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Robert Haas <robertmhaas(at)gmail(dot)com> writes:
>> On Wed, Sep 9, 2015 at 11:49 PM, Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
>>> - errdetail("Could not rename \"%s\" to \"%s\": %m.",
>>> + errdetail("\"%s\" could not be renamed to \"%s\": %m.",
>>>
>>> Is there any reason to change this message?
>>> I think you have changed this message to make it somewhat similar with
>>> the new message we are planning to use in this function, but I don't see
>>> that as compelling reason to change this message.
>
>> The old message better follows the guidelines. See section 51.3.7:
>> Avoid Passive Voice. The old message is what's called
>> "telegram-style", with PostgreSQL itself as the implicit subject. The
>> proposed replacement is just the regular old passive voice.
>
> Neither version is following the guidelines very well, in particular they
> should be mentioning what kind of object %s is (file? table? tablespace?).
> But to me the "could not be renamed" version seems to be closer to the
> spirit of the "use complete sentences" rule for errdetail. The other one
> seems better fit for a primary error message, which is supposed to be
> kept short.

Hmm, I did miss the fact that this was an errdetail(). I agree that
the object type should be mentioned either way. Actually, it's sort
of surprising that this message is a detail message rather than a
primary message.

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company