Обсуждение: BUG #18034: Accept the spelling "+infinity" in datetime input is not accurate

The following bug has been logged on the website:

Bug reference:      18034
Logged by:          yanliang lei
Email address:      msdnchina@163.com
PostgreSQL version: 16beta2
Operating system:   RHEL7.9
Description:

in the https://www.postgresql.org/docs/16/release-16.html, there is a new
feature ：Accept the spelling "+infinity" in datetime input。
Actually，"in datetime input " is not accurate.
Actually, in the PostgreSQL 16 beta2 edition, the spelling "+infinity" can
be used in the following three datatypes:
1.timestamp [ (p) ] [ without time zone ]
2.timestamp [ (p) ] with time zone
3.date

and the spelling "+infinity" can not be used in the   following two
datatypes:
1.time [ (p) ] [ without time zone ]
2.time [ (p) ] with time zone

so ，I think that the follow descirtion is accurate：
Accept the spelling "+infinity" in the following data types:
      timestamp [ (p) ] [ without time zone ] 、
      timestamp [ (p) ] with time zone、
      date datatypes。

On Thu, Jul 20, 2023 at 03:25:11PM +0000, PG Bug reporting form wrote:
> The following bug has been logged on the website:
> 
> Bug reference:      18034
> Logged by:          yanliang lei
> Email address:      msdnchina@163.com
> PostgreSQL version: 16beta2
> Operating system:   RHEL7.9
> Description:        
> 
> in the https://www.postgresql.org/docs/16/release-16.html, there is a new
> feature ：Accept the spelling "+infinity" in datetime input。
> Actually，"in datetime input " is not accurate.
> Actually, in the PostgreSQL 16 beta2 edition, the spelling "+infinity" can
> be used in the following three datatypes:
> 1.timestamp [ (p) ] [ without time zone ]
> 2.timestamp [ (p) ] with time zone
> 3.date
> 
> and the spelling "+infinity" can not be used in the   following two
> datatypes:
> 1.time [ (p) ] [ without time zone ]
> 2.time [ (p) ] with time zone
> 
> so ，I think that the follow descirtion is accurate：
> Accept the spelling "+infinity" in the following data types:
>       timestamp [ (p) ] [ without time zone ] 、
>       timestamp [ (p) ] with time zone、
>       date datatypes。

We call our timestamp type datetime in some cases, e.g.:

    https://www.postgresql.org/docs/16/runtime-config-client.html#RUNTIME-CONFIG-CLIENT-FORMAT
    
    timezone_abbreviations (string)
    
        Sets the collection of time zone abbreviations that will be accepted
        by the server for datetime input.

I see it in a few other places.  Should we rename it other places too? 
I thought datetime was just a short-hand for our date-time types.

-- 
  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
  EDB                                      https://enterprisedb.com

  Only you can decide what is important to you.

Bruce Momjian <bruce@momjian.us> writes:
>> in the https://www.postgresql.org/docs/16/release-16.html, there is a new
>> feature ：Accept the spelling "+infinity" in datetime input。
>> Actually，"in datetime input " is not accurate.
>> Actually, in the PostgreSQL 16 beta2 edition, the spelling "+infinity" can
>> be used in the following three datatypes:
>> 1.timestamp [ (p) ] [ without time zone ]
>> 2.timestamp [ (p) ] with time zone
>> 3.date
>> 
>> and the spelling "+infinity" can not be used in the   following two
>> datatypes:
>> 1.time [ (p) ] [ without time zone ]
>> 2.time [ (p) ] with time zone

> We call our timestamp type datetime in some cases, e.g.:
> ...
> I see it in a few other places.  Should we rename it other places too? 
> I thought datetime was just a short-hand for our date-time types.

I don't see much reason to change anything here.  "Datetime" is not
a perfectly strict classification, eg it might or might not include
"interval" depending on context, and I don't want to try to make
that exact.

A more specific release note entry could be "Accept the spelling
'+infinity' for datetime types that accept infinity"; but I'm not
sure it's worth the extra verbiage.

            regards, tom lane

On Wed, Aug  9, 2023 at 10:53:27PM -0400, Tom Lane wrote:
> Bruce Momjian <bruce@momjian.us> writes:
> > We call our timestamp type datetime in some cases, e.g.:
> > ...
> > I see it in a few other places.  Should we rename it other places too? 
> > I thought datetime was just a short-hand for our date-time types.
> 
> I don't see much reason to change anything here.  "Datetime" is not
> a perfectly strict classification, eg it might or might not include
> "interval" depending on context, and I don't want to try to make
> that exact.
> 
> A more specific release note entry could be "Accept the spelling
> '+infinity' for datetime types that accept infinity"; but I'm not
> sure it's worth the extra verbiage.

Yeah, that was my analysis too.

-- 
  Bruce Momjian  <bruce@momjian.us>        https://momjian.us
  EDB                                      https://enterprisedb.com

  Only you can decide what is important to you.

On 8/18/23 15:57, yanliang lei wrote:
> Dear everyone，
> I raise this question based on the following considerations:
> 1. Make PostgreSQL easier for beginners to learn；
> 2. PostgreSQL is a database software product rather than a database software project. A product means being more
user-friendly,precise, and reducing the likelihood of misunderstandings。
 
> 
> 
> The datetime type may make beginners think that there is a datetime data type in the PostgreSQL database. Although
seniorPostgreSQL professionals know that there is no datetime data type, the datetime type is enough to make beginners
misunderstand.
> However, When they saw that the ReleaseNote in PostgreSQL 16 recorded "Accept the spelling"+infinity "in datetime
typeinput"，
 
>   the term 'datetime type' is sufficient for beginners to misunderstand , unless it is noted that 'datetime type' is
ageneral term, representing the following PostgreSQL data types: timestamp [(p)] [without time zone], timestamp [(p)]
withtime zone, date。
 
> 
> 
> I am not nitpicking ，I hope that: PostgreSQL database software products can become more perfect and accurate。
> Thanks a lot！

On the other hand, the standard talks about datetime types as a 
collection of dates, times, timestamps, and intervals; and there is no 
type called "datetime" in the standard.  We are just doing the same 
thing here.
-- 
Vik Fearing