On Mon, Oct 18, 2010 at 1:07 PM, Pavel Stehule <pavel.stehule@gmail.com> wrote:
> 2010/10/18, Robert Haas <robertmhaas@gmail.com>:
>> On Mon, Oct 18, 2010 at 7:37 AM, Itagaki Takahiro
>> <itagaki.takahiro@gmail.com> wrote:
>>> On Sat, Oct 16, 2010 at 7:29 AM, Robert Haas <robertmhaas@gmail.com>
>>> wrote:
>>>> No doubt. The problem is that we're going to end up with those bells
>>>> and whistles in two places: in to_char or other type-specific
>>>> formatting functions, and again in format.
>>>
>>> If we decide to use C-like sprintf(), I think the only thing we can do
>>> is to implement C-syntax as much as possible. Users will expect the
>>> function behaves as sprintf, because it has the similar syntax.
>>> It's not an item for now, but someone would request it at a future date.
>>>
>>>
>>> BTW, the interoperability is why I proposed {} syntax. For example,
>>> {1:YYYY-MM-DD} for date is expanded to to_char($1, 'YYYY-MM-DD').
>>> (Maybe it's not so easy; It requires function lookups depending on types.)
>>
>> There's no particular reason why we couldn't make this work with
>> sprintf-type syntax; for example, you could allow %{XYZ} to mean
>> to_char(value, 'XYZ'). But it seems to me that we have agreement that
>> this should start with just %s, %I, %L and allow 3$ or similar in the
>> middle to specify which argument it is. We can then argue about how
>> many more bells and whistles to add later.
>
> so, yes. Can we finish this discus with this result? I'll prepare
> patch for next commit fest. Next question - what about sprintf
> function in core? Is living this idea still?
I'm indifferent about whether we put it in core or contrib.
--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company