Time object problems - 1.8.2

ruby-talk
1

I am trying to efficiently get the date part of a Time object.

To remove the time part of the Time object I subtract the modulo of a
day-in-seconds. But if I do this I get a Time object of the date *plus
1 hour*!!! Perhaps I'm just tired, but I don;t understand why.

ie :

  DAY_IN_SECS=60*60*24;
  t=Time.parse("20-Oct-2005 04:04:00")
  datePart=t-t.to_f.modulo(DAY_IN_SECS);

(and I might apply Time.at to get a Time object rather than a float.)

However, the result of the above is

  20-Oct-2005 01:00:00

instead of

  20-Oct-2005 00:00:00

Anyone got any ideas?

Greg

2

Greg Lorriman scribbled on Wednesday 15 Mar 2006 12:55:

I am trying to efficiently get the date part of a Time object.

To remove the time part of the Time object I subtract the modulo of a
day-in-seconds. But if I do this I get a Time object of the date *plus
1 hour*!!! Perhaps I'm just tired, but I don;t understand why.

ie :

  DAY_IN_SECS=60*60*24;
  t=Time.parse("20-Oct-2005 04:04:00")
  datePart=t-t.to_f.modulo(DAY_IN_SECS);

(and I might apply Time.at to get a Time object rather than a float.)

However, the result of the above is

  20-Oct-2005 01:00:00

instead of

  20-Oct-2005 00:00:00

Anyone got any ideas?

Greg

irb(main):001:0> t = Time.now; t - t.sec - 60 * t.min - 3600 * t.hour
=> Wed Mar 15 00:00:00 CET 2006

3

Greg Lorriman schrieb:

I am trying to efficiently get the date part of a Time object.

To remove the time part of the Time object I subtract the modulo of a
day-in-seconds. But if I do this I get a Time object of the date *plus
1 hour*!!! Perhaps I'm just tired, but I don;t understand why.

  DAY_IN_SECS=60*60*24;
  t=Time.parse("20-Oct-2005 04:04:00")
  datePart=t-t.to_f.modulo(DAY_IN_SECS);

(and I might apply Time.at to get a Time object rather than a float.)

However, the result of the above is

  20-Oct-2005 01:00:00

instead of

  20-Oct-2005 00:00:00

Anyone got any ideas?

Greg, look at this:

   p Time.at(0)

It starts at 1 o'clock.

Regards,
Pit

4

Hi Greg,

I am trying to efficiently get the date part of a Time object.

I've just been playing around with Time and Date a bit. I have to ask: why?

You can get just the date part much more easily by using a Date object:

Date.parse("20-Oct-2005 04:04:00").to_s
=> "2005-10-20"

If you absolutely need a time object out of it, you can use 'facets':

require 'facet/date/to_time'
Date.parse("20-Oct-2005 04:04:00").to_time
=> Thu Oct 20 00:00:00 CEST 2005

HTH

- Dimitri

···

On 3/15/06, Greg Lorriman <temp@lorriman.com> wrote:

5

I've just been playing around with Time and Date a bit. I have to ask: why?

You can get just the date part much more easily by using a Date object:

Unfortunately the Date class is horrifically slow. I'm already being
hammered by other slow aspects of ruby, which I can't avoid. So Time it
is!

6

irb(main):001:0> t = Time.now; t - t.sec - 60 * t.min - 3600 * t.hour
=> Wed Mar 15 00:00:00 CET 2006

How about:

irb(main):001:0> t = Time.now; Time.local(t.year,t.month,t.day)
=> Wed Mar 15 00:00:00 Central Standard Time 2006

7

Anyone got any ideas?

Greg, look at this:

  p Time.at(0)

It starts at 1 o'clock.

Unfathomable! What does it mean?!?! I am entirely perplexed by this
reality.

Greg

8

Indeed. I just discovered
http://www.recentrambles.com/pragmatic/view/33 from another thread.
Interestingly enough, by replacing DateTime with Date in the example,
the benchmark comes out about 3 times faster. I guess having to work
with hours, minutes, and seconds is expensive.

- Dimitri

···

On 3/15/06, Greg Lorriman <temp@lorriman.com> wrote:

Unfortunately the Date class is horrifically slow. I'm already being
hammered by other slow aspects of ruby, which I can't avoid. So Time it
is!

9

Greg Lorriman schrieb:

p Time.at(0)
It starts at 1 o'clock.

Unfathomable! What does it mean?!?! I am entirely perplexed by this
reality.

Greg, sorry for the noise. Time.at(0) is "epoch" (1970-01-01 00:00:00 GMT). My timezone is GMT + 1, so epoch happened to be at 1 o'clock. I suppose that's the reason why your algorithm didn't work.

Regards,
Pit

10

It makes me think that an effort is needed to convert many of the
libraries that do basic heavy-lifting, like DateTime and CSV and
others, to C libraries.

I like the purity of the idea of a "pure Ruby" library, but I am
suffering the consequences and even considering jumping to Python; the
pain is getting to be too much.

11

p Time.at(0)
It starts at 1 o'clock.

Unfathomable! What does it mean?!?! I am entirely perplexed by this
reality.

Greg, sorry for the noise. Time.at(0) is "epoch" (1970-01-01 00:00:00
GMT). My timezone is GMT + 1, so epoch happened to be at 1 o'clock. I
suppose that's the reason why your algorithm didn't work.

Well that is the answer, then. Thankyou very much for that. Perhaps the
docs need expanding a little. I think even a computer scientist might
trip up on "epoch". I was thinking, in fact, that Time isn't a
practical Time/Date implementation and was possibly concieved in an
academic mind, which your observation would seem to confirm; I don't
think I've ever used such an awkward time/date implementation. However
I haven't used Date yet!!! :slight_smile:

Greg

12

check out the speed gains one can get with FasterCSV and reconsider - and it's
still pure ruby.

regards.

-a

···

On Thu, 16 Mar 2006, Greg Lorriman wrote:

It makes me think that an effort is needed to convert many of the
libraries that do basic heavy-lifting, like DateTime and CSV and
others, to C libraries.

I like the purity of the idea of a "pure Ruby" library, but I am
suffering the consequences and even considering jumping to Python; the
pain is getting to be too much.

--
share your knowledge. it's a way to achieve immortality.
- h.h. the 14th dali lama

13

I like the purity of the idea of a "pure Ruby" library, but I am
suffering the consequences and even considering jumping to Python; the
pain is getting to be too much.

check out the speed gains one can get with FasterCSV and reconsider - and it's
still pure ruby

Sounds promising. I particularly like the sound of "Faster", in
"FasterCSV".

thanks for that

Greg