|
|
|
|
| Re: linfit with string time array (x) [message #63430 is a reply to message #63426] |
Tue, 11 November 2008 00:51   |
R.Bauer
Messages: 1424
Registered: November 1998
|
Senior Member |
|
|
julia.walterspiel@gmail.com schrieb:
> On 11 Nov., 06:45, l...@lbnc.de wrote:
>> On 10 Nov, 16:43, Reimar Bauer <R.Ba...@fz-juelich.de> wrote:
>>
>>
>>
>>> if the numbers are written in a time format it is better to show an
>>> example. Those need to be converted into a number. I would suggest to
>>> use julian seconds (seconds since 2000-01-01 00:00:00 UTC)
>> But why?! Why go through the bother when IDL has JULDAY and CALDAT?
>> With the LABEL_DATE function you get nice annotations of axes and you
>> don't need to worry about whether the routines you download somewhere
>> of the web are actually right.
>>
>> To answer the original question, as Reimar alrady said, you need to
>> convert the strings to some sort of numeric value, for example julian
>> day. For that you need to parse the year, month, day, hour, minute,
>> second from your string and pass the to, for example, JULDAY. The IDL
>> documentation knows more about how to use JULDAY and LABEL_DATE.
>>
>> Cheers
>> Lasse Clausen
>
> hi guys
> well I thought I had already converted the string to some sort of
> numeric value but I guess it was just the problem of doing what Reimar
> said (new = double(date)). here's how I extracted my date from the
> file name:
>
> date = STRMID(filename_short, 10,7)
> year = Fix(StrMid(StrTrim(date,2), 0, 4))
> dayofyear = Fix(StrMid(StrTrim(date,2), 4, 3))
> CALDAT, JULDAY(1, dayofyear, year), month, day
> date = julday(month, day, year)
>
> --> then my date of data looks like this: 2451605
IDL> date = julday(1, 1, 2008)
% Compiled module: JULDAY.
IDL> print,date
2454467
IDL> help,date
DATE LONG = 2454467
If you use help you get the type of the var. It is of type long not string.
If you haven't read the article
http://www.dfanning.com/math_tips/sky_is_falling.html
it is quite interesting to see how a computer knows a number.
>
> That's what confuses me; a number that is obviously considered a
> string. so I guess IDL doesn't take the "number" but rather what is
> behind the calculation (julday) and refers to that as a string? Or am
> I misunderstanding something?
> It works fine when I convert it to double, so thanks for that input. I
you are welcome
cheers
Reimar
|
|
|
|
| Re: linfit with string time array (x) [message #63431 is a reply to message #63430] |
Tue, 11 November 2008 00:38 |
R.Bauer
Messages: 1424
Registered: November 1998
|
Senior Member |
|
|
lbnc@lbnc.de schrieb:
> On 10 Nov, 16:43, Reimar Bauer <R.Ba...@fz-juelich.de> wrote:
>> if the numbers are written in a time format it is better to show an
>> example. Those need to be converted into a number. I would suggest to
>> use julian seconds (seconds since 2000-01-01 00:00:00 UTC)
>
> But why?! Why go through the bother when IDL has JULDAY and CALDAT?
> With the LABEL_DATE function you get nice annotations of axes and you
> don't need to worry about whether the routines you download somewhere
> of the web are actually right.
Well you have to write your own tests for nearly everything you use or
program by your own (or have been by others). A simple doctest example
in the code does help a lot. (idl has no unittest framework)
Sometimes you should use a calculator or/and look into some book of
tables. And if you look into books it is better to not believe only on
one book. ;)
I never have seen any unit tests by idl developing team in the last 15
years. But I have send them some test examples for every bug I found. So
we can assume they do unittests. But they are not part of the product.
Sometimes it is better to use your own well tested routine. ;)
>
> To answer the original question, as Reimar alrady said, you need to
> convert the strings to some sort of numeric value, for example julian
> day. For that you need to parse the year, month, day, hour, minute,
> second from your string and pass the to, for example, JULDAY. The IDL
> documentation knows more about how to use JULDAY and LABEL_DATE.
If you like julday just use julday. :)
cheers
Reimar
|
|
|
|
| Re: linfit with string time array (x) [message #63432 is a reply to message #63431] |
Tue, 11 November 2008 00:28 |
julia.walterspiel
Messages: 35
Registered: July 2008
|
Member |
|
|
On 11 Nov., 06:45, l...@lbnc.de wrote:
> On 10 Nov, 16:43, Reimar Bauer <R.Ba...@fz-juelich.de> wrote:
>
>
>
>> if the numbers are written in a time format it is better to show an
>> example. Those need to be converted into a number. I would suggest to
>> use julian seconds (seconds since 2000-01-01 00:00:00 UTC)
>
> But why?! Why go through the bother when IDL has JULDAY and CALDAT?
> With the LABEL_DATE function you get nice annotations of axes and you
> don't need to worry about whether the routines you download somewhere
> of the web are actually right.
>
> To answer the original question, as Reimar alrady said, you need to
> convert the strings to some sort of numeric value, for example julian
> day. For that you need to parse the year, month, day, hour, minute,
> second from your string and pass the to, for example, JULDAY. The IDL
> documentation knows more about how to use JULDAY and LABEL_DATE.
>
> Cheers
> Lasse Clausen
hi guys
well I thought I had already converted the string to some sort of
numeric value but I guess it was just the problem of doing what Reimar
said (new = double(date)). here's how I extracted my date from the
file name:
date = STRMID(filename_short, 10,7)
year = Fix(StrMid(StrTrim(date,2), 0, 4))
dayofyear = Fix(StrMid(StrTrim(date,2), 4, 3))
CALDAT, JULDAY(1, dayofyear, year), month, day
date = julday(month, day, year)
--> then my date of data looks like this: 2451605
That's what confuses me; a number that is obviously considered a
string. so I guess IDL doesn't take the "number" but rather what is
behind the calculation (julday) and refers to that as a string? Or am
I misunderstanding something?
It works fine when I convert it to double, so thanks for that input. I
|
|
|
|
| Re: linfit with string time array (x) [message #63434 is a reply to message #63432] |
Mon, 10 November 2008 21:45 |
lbnc
Messages: 15
Registered: January 2005
|
Junior Member |
|
|
On 10 Nov, 16:43, Reimar Bauer <R.Ba...@fz-juelich.de> wrote:
>
> if the numbers are written in a time format it is better to show an
> example. Those need to be converted into a number. I would suggest to
> use julian seconds (seconds since 2000-01-01 00:00:00 UTC)
But why?! Why go through the bother when IDL has JULDAY and CALDAT?
With the LABEL_DATE function you get nice annotations of axes and you
don't need to worry about whether the routines you download somewhere
of the web are actually right.
To answer the original question, as Reimar alrady said, you need to
convert the strings to some sort of numeric value, for example julian
day. For that you need to parse the year, month, day, hour, minute,
second from your string and pass the to, for example, JULDAY. The IDL
documentation knows more about how to use JULDAY and LABEL_DATE.
Cheers
Lasse Clausen
|
|
|
|
|
|
| Re: linfit with string time array (x) [message #63558 is a reply to message #63431] |
Tue, 11 November 2008 12:31 |
Michael Galloy
Messages: 1114
Registered: April 2006
|
Senior Member |
|
|
On Nov 11, 1:38 am, Reimar Bauer <R.Ba...@fz-juelich.de> wrote:
> program by your own (or have been by others). A simple doctest example
> in the code does help a lot. (idl has no unittest framework)
Nothing part of IDL's library, but I do have a unit testing framework:
http://michaelgalloy.com/2007/07/03/update-to-mgunit.html
By the way, do you have an automated way to run the doctest examples
you are talking about? I have thought about adding something like that
to IDLdoc.
Mike
--
www.michaelgalloy.com
Tech-X Corporation
Associate Research Scientist
|
|
|
|
comp.lang.idl-pvwave archive
Members
Search
Help
Login
Home
