Postgres Pro
Downloads
Home   >   mailing lists

Thread: timestamp typedefs

timestamp typedefs

From
"Warren Turkal"
Date:
Hello,

I have created the following patch in an effort to start cleaning up
the timestamp datatype. Please let me know if something like this will
help so that I know whether to keep going. BTW, it passes a "make
check" AFAICT.

Thanks,
wt

Re: timestamp typedefs

From
Tom Lane
Date:
"Warren Turkal" <wturkal@gmail.com> writes:
> I have created the following patch in an effort to start cleaning up
> the timestamp datatype. Please let me know if something like this will
> help so that I know whether to keep going. BTW, it passes a "make
> check" AFAICT.

Do we really need "fhour_t" and "fminute_t" on top of "fsec_t"?
This seems like a bad factorization ...
        regards, tom lane

Re: timestamp typedefs

From
Tom Lane
Date:
I wrote:
> Do we really need "fhour_t" and "fminute_t" on top of "fsec_t"?
> This seems like a bad factorization ...

After some more thought: I think that what's bugging me is that "fsec_t"
is intended to denote "fractional seconds".  The other cases you have
here seem not to be intended to be "fractional hours" or "fractional
minutes".  I'm not quite sure what the right abstraction is, but it
doesn't seem to be that.
        regards, tom lane

Re: timestamp typedefs

From
"Warren Turkal"
Date:
On Jan 3, 2008 8:54 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> I wrote:
> > Do we really need "fhour_t" and "fminute_t" on top of "fsec_t"?
> > This seems like a bad factorization ...
>
> After some more thought: I think that what's bugging me is that "fsec_t"
> is intended to denote "fractional seconds".  The other cases you have
> here seem not to be intended to be "fractional hours" or "fractional
> minutes".  I'm not quite sure what the right abstraction is, but it
> doesn't seem to be that.

I thought it meant "field seconds". That's why I used fhour_t and
fminute_t. I'll think about a better name.

wt

Re: timestamp typedefs

From
Alvaro Herrera
Date:
Warren Turkal escribió:
> On Jan 3, 2008 8:54 PM, Tom Lane <tgl@sss.pgh.pa.us> wrote:
> > I wrote:
> > > Do we really need "fhour_t" and "fminute_t" on top of "fsec_t"?
> > > This seems like a bad factorization ...
> >
> > After some more thought: I think that what's bugging me is that "fsec_t"
> > is intended to denote "fractional seconds".  The other cases you have
> > here seem not to be intended to be "fractional hours" or "fractional
> > minutes".  I'm not quite sure what the right abstraction is, but it
> > doesn't seem to be that.
> 
> I thought it meant "field seconds". That's why I used fhour_t and
> fminute_t. I'll think about a better name.

Perhaps what you want here is to define a type for calculation results
(double/int64).  Whether it is used in the code for minutes or hours is
irrelevant to the typedef.

-- 
Alvaro Herrera                                http://www.CommandPrompt.com/
The PostgreSQL Company - Command Prompt, Inc.

Re: timestamp typedefs

From
"Warren Turkal"
Date:
On Jan 4, 2008 4:20 AM, Alvaro Herrera <alvherre@commandprompt.com> wrote:
> Perhaps what you want here is to define a type for calculation results
> (double/int64).  Whether it is used in the code for minutes or hours is
> irrelevant to the typedef.

Okay...that sounds good. Do you have a good name for it?
Alternatively, we could just use a TimestampTZ, I guess.

wt