Re: proposal: minscale, rtrim, btrim functions for numeric

ne 10. 11. 2019 v 7:35 odesílatel Pavel Stehule <pavel.stehule@gmail.com>
napsal:

>
>
> so 9. 11. 2019 v 21:34 odesílatel Tom Lane <tgl@sss.pgh.pa.us> napsal:
>
>> Pavel Stehule <pavel.stehule@gmail.com> writes:
>> > four years ago Marko Tiikkaja send a patch for numeric_trim functions.
>> This
>> > functions removed ending zeroes from numeric value. This is useful
>> feature,
>> > but there was not any progress on this patch. I think so this feature
>> can
>> > be interesting, so I would to revitalize this patch.
>>
>> > Original discussion
>> >
>> https://www.postgresql-archive.org/Add-numeric-trim-numeric-td5874444.html
>>
>> A more useful link is
>> https://www.postgresql.org/message-id/flat/564D3ADB.7000808%40joh.to
>> and the earlier discussion is at
>> https://www.postgresql.org/message-id/flat/5643125E.1030605%40joh.to
>>
>> Re-reading that thread, I don't really think there's much support for
>> anything beyond the minscale() function.  The rest are just inviting
>> confusion with string-related functions.  And I really don't like
>> establishing a precedent that btrim() and rtrim() are the same.
>>
>
> I have to agree, so using trim, rtrim names is not best. On second hand,
> probably to most often usage of minscale function will be inside expression
> round(x, minscale(x)), so this functionality can be in core. A question is
> a name.
>
> maybe to_minscale(numeric) ?
>

Here is a patch. It's based on Dean's suggestions.

I implemented two functions - first minscale, second trim_scale. The
overhead of second is minimal - so I think it can be good to have it. I
started design with the name "trim_scale", but the name can be any other.

Regards

Pavel


>
> Regards
>
> Pavel
>
>
>>                         regards, tom lane
>>
>