Here's a special thanks to BabakNYC, David Briars, J.Gross, and Hallway for their great input.
I'm seeing an issue with the DTPART function in 8204. Is anyone else seeing such symptoms?
Here's my pseudo-code:
TABLE FILE DATE_TO_CHECK
"DTPART Check for Week in Date"
SUM
COMPUTE YEAR/I4 = DTPART(DATE_TO_CHECK, YEAR) ;
COMPUTE WEEK/I2 = DTPART(DATE_TO_CHECK, WEEK) ;
BY DATE_TO_CHECK AS 'Date To Check'
WHERE DATE_TO_CHECK FROM '20171221' TO '20180131'
ON TABLE PCHOLD FORMAT XLSX
END
and, here's my results: Why is January 1 2018 in Week 53?
DTPART Check for Week in Date
Date To Check YEAR WEEK
2017/12/21 2017 51
2017/12/23 2017 51
2017/12/24 2017 52
2017/12/30 2017 52
2018/01/01 2018 53 <---
2018/01/05 2018 53 <---
2018/01/07 2018 1
2018/01/13 2018 1
2018/01/14 2018 2
2018/01/20 2018 2
2018/01/21 2018 3
2018/01/27 2018 3
2018/01/28 2018 4
2018/01/29 2018 4
Any suggestions on how to alleviate this are graciously welcomed. Thank You, DougThis message has been edited. Last edited by: Doug,
In FOCUS Since 1983 ~ from FOCUS to WebFOCUS. Current: WebFOCUS Administrator at FIS Worldpay | 8204, 8206
October 31, 2018, 12:35 PM
Doug
Sorry... I often ask "Can you do that with the CAR file?". So, here it is:
With these results (The "<--" were added for clarity):
DTPART Test
20171228 52 Day 5
20171229 52 Day 6
20171230 52 Day 7
20171231 53 Day 1
20180101 53 Day 2 <--
20180102 53 Day 3 <--
20180103 53 Day 4 <--
20180104 53 Day 5 <--
20180105 53 Day 6 <--
20180106 53 Day 7 <--
20180107 1 Day 1
20180108 1 Day 2
It looks like Week 53 is a carry-over from 2017. Should that be?This message has been edited. Last edited by: Doug,
October 31, 2018, 01:06 PM
BabakNYC
Doug: I looked this concept up and found an article that basically says:
"While it only happens every five or six years, a 53-week broadcast year isn’t unheard of. These happen around a leap year or when January 1 falls on a Saturday or Sunday. That means 2017 is a 53-week year, which hasn’t happened since 2012. In 2017, 4th quarter will have 14 weeks with October and December each being 5 weeks long. Additional years where there will be 53 weeks include 2023, 2028 and 2034."
WebFOCUS 8206, Unix, Windows
October 31, 2018, 01:12 PM
David Briars
Can you do a '? SET WEEKFIRST' to see the value of WEEKFIRST at your site?
SET WEEKFIRST = 2
-RUN
DEFINE FILE CAR
DATE_TO_CHECK/HYYMD = DT(20180101);
WEEK/I2 = DTPART(DATE_TO_CHECK, WEEK) ;
END
TABLE FILE CAR
PRINT CAR DATE_TO_CHECK WEEK
IF RECORDLIMIT EQ 1
END
-EXIT
Shows me week 1, and
SET WEEKFIRST = 1
-RUN
DEFINE FILE CAR
DATE_TO_CHECK/HYYMD = DT(20180101);
WEEK/I2 = DTPART(DATE_TO_CHECK, WEEK) ;
END
TABLE FILE CAR
PRINT CAR DATE_TO_CHECK WEEK
IF RECORDLIMIT EQ 1
END
-EXIT
Shows week 53.
Pilot: WebFOCUS 8.2.06 Test: WebFOCUS 8.1.05M Prod: WebFOCUS 8.1.05M Server: Windows Server 2016/Tomcat Standalone Workstation: Windows 10/IE11+Edge Database: Oracle 12c, Netezza, & MS SQL Server 2019 Output: AHTML/XLSX/HTML/PDF/JSCHART Tools: WFDS, Repository Content, BI Portal Designer & ReportCaster
October 31, 2018, 01:13 PM
Doug
Thanks, I understand that there are "53 week years". But, is the week always complete weeks, based on the first day of the week? And, if so, it can be different based on what the first day of the week is, right?
Here's the underlying issue: I'm converting reports from Tableau to WebFOCUS based on week. WebFOCUS is showing the weeks being one week behind that shown in the Tableau reports. BTW: Outlook even says we're currently in week 44, while WebFOCUS says we're in week 43. I'm checking with my Tableau team on this as well.
October 31, 2018, 01:21 PM
BabakNYC
I haven't done much with this. I'd say this is worth a conversation with IB Tech support. If Outlook is saying something different from what WebFOCUS generates then I'd love to know IB's explanation.
WebFOCUS 8206, Unix, Windows
October 31, 2018, 01:28 PM
Doug
Excellent point David! ? SET WEEKFIRST It shows as "WEEKFIRST 0" However, I ran that same report with the following code prepended to it, and got the same results.
? SET WEEKFIRST
-RUN
SET WEEKFIRST = 1
-RUN
? SET WEEKFIRST
-EXIT
The final result of the "? SET WEEKFIRST" shows it as 1. Could it be a DTPART routine issue not recognizing the setting of weekfirst?
October 31, 2018, 01:30 PM
Doug
quote:
I'd love to know IB's explanation.
BabakNYC ... Me too ... Even with the SET WEEKFIRST ...
It's time for a case.
October 31, 2018, 01:33 PM
Doug
Can anyone get anything else from the code in my posted Wed Oct 31 2018 12:35 post, especially using 8204?
In FOCUS Since 1983 ~ from FOCUS to WebFOCUS. Current: WebFOCUS Administrator at FIS Worldpay | 8204, 8206
October 31, 2018, 02:06 PM
j.gross
The Keysheet lists the options for SET WEEKFIRST=option as option: 1 to 7 (Sunday to Saturday) option: ISO2
Have you tried the ISO2 option?
-Jack
October 31, 2018, 02:41 PM
Hallway
quote:
Originally posted by Doug: Can anyone get anything else from the code in my posted Wed Oct 31 2018 12:35 post, especially using 8204?
Try this:
SET WEEKFIRST = ISO1
TABLE FILE CAR
"DTPART Test"
SUM DCOST NOPRINT
COMPUTE 20171228/I2 = DTPART('20171228', WEEK) ; COMPUTE Day/I2 = DTPART('20171228', WEEKDAY) ; OVER
COMPUTE 20171229/I2 = DTPART('20171229', WEEK) ; COMPUTE Day/I2 = DTPART('20171229', WEEKDAY) ; OVER
COMPUTE 20171230/I2 = DTPART('20171230', WEEK) ; COMPUTE Day/I2 = DTPART('20171230', WEEKDAY) ; OVER
COMPUTE 20171231/I2 = DTPART('20171231', WEEK) ; COMPUTE Day/I2 = DTPART('20171231', WEEKDAY) ; OVER
COMPUTE 20180101/I2 = DTPART('20180101', WEEK) ; COMPUTE Day/I2 = DTPART('20180101', WEEKDAY) ; OVER
COMPUTE 20180102/I2 = DTPART('20180102', WEEK) ; COMPUTE Day/I2 = DTPART('20180102', WEEKDAY) ; OVER
COMPUTE 20180103/I2 = DTPART('20180103', WEEK) ; COMPUTE Day/I2 = DTPART('20180103', WEEKDAY) ; OVER
COMPUTE 20180104/I2 = DTPART('20180104', WEEK) ; COMPUTE Day/I2 = DTPART('20180104', WEEKDAY) ; OVER
COMPUTE 20180105/I2 = DTPART('20180105', WEEK) ; COMPUTE Day/I2 = DTPART('20180105', WEEKDAY) ; OVER
COMPUTE 20180106/I2 = DTPART('20180106', WEEK) ; COMPUTE Day/I2 = DTPART('20180106', WEEKDAY) ; OVER
COMPUTE 20180107/I2 = DTPART('20180107', WEEK) ; COMPUTE Day/I2 = DTPART('20180107', WEEKDAY) ; OVER
COMPUTE 20180108/I2 = DTPART('20180108', WEEK) ; COMPUTE Day/I2 = DTPART('20180108', WEEKDAY) ;
BY COUNTRY NOPRINT
WHERE RECORDLIMIT EQ 1
END
Below is a query to see all the available values of WEEKFIRST:
TABLE FILE sysset
BY SETNAME
BY SETDESC
BY SERVDESC
BY SETVALTYPE
BY SETMAXLEN AS 'Max,Length/,Value'
BY SETFPROMPT AS 'Set,From,PROMPT'
BY SETFTAB AS 'Set,From,TABLE'
BY SETFPARMS AS 'Set,From,PARAMS'
BY CURR_VALUE
BY VALUE AS 'Available,Values'
BY VALDESC
BY IS_DEFAULT
WHERE SETNAME EQ 'WEEKFIRST';
ON TABLE SET PAGE-NUM OFF
END
Hallway
Prod: 8202M1
Test: 8202M4
Repository:
OS:
Outputs:
October 31, 2018, 03:26 PM
Doug
Thanks Guys. It looks like WEEKFIRST = ISO2 is the winner, I need Jan 1 to be the start of week 1. Here's my results based on the great info from J.Gross and Hallway:
DTPART Test: WEEKFIRST = 1 DTPART Test: WEEKFIRST = ISO1 DTPART Test: WEEKFIRST = ISO2
20171228 52 Day 5 20171228 52 Day 5 20171228 52 Day 4
20171229 52 Day 6 20171229 52 Day 6 20171229 52 Day 5
20171230 52 Day 7 20171230 52 Day 7 20171230 52 Day 6
20171231 53 Day 1 20171231 1 Day 1 20171231 52 Day 7
20180101 53 Day 2 20180101 1 Day 2 20180101 1 Day 1
20180102 53 Day 3 20180102 1 Day 3 20180102 1 Day 2
20180103 53 Day 4 20180103 1 Day 4 20180103 1 Day 3
20180104 53 Day 5 20180104 1 Day 5 20180104 1 Day 4
20180105 53 Day 6 20180105 1 Day 6 20180105 1 Day 5
20180106 53 Day 7 20180106 1 Day 7 20180106 1 Day 6
20180107 1 Day 1 20180107 2 Day 1 20180107 1 Day 7
20180108 1 Day 2 20180108 2 Day 2 20180108 2 Day 1
20181026 42 Day 6 20181026 43 Day 6 20181026 43 Day 5
20181027 42 Day 7 20181027 43 Day 7 20181027 43 Day 6
20181028 43 Day 1 20181028 44 Day 1 20181028 43 Day 7
20181029 43 Day 2 20181029 44 Day 2 20181029 44 Day 1
20181030 43 Day 3 20181030 44 Day 3 20181030 44 Day 2
20181031 43 Day 4 20181031 44 Day 4 20181031 44 Day 3
SET WEEKFIRST: Start of week for date/datetime functions' WEEK and WEEKDAY components.
DEFAULT 1 ... Current Value: ISO2
Values Available Value Description
0 No setting.
1 Sunday, non-standard week numbering.
2 Monday, non-standard week numbering.
3 Tuesday, non-standard week numbering.
4 Wednesday, non-standard week numbering.
5 Thursday, non-standard week numbering.
6 Friday, non-standard week numbering.
7 Saturday, non-standard week numbering.
ISO ISO standard establishes Monday as 1st day of week, so allow alternate of ISO for ISO2.
ISO1 Sunday, ISO standard week numbering.
ISO2 Monday, ISO standard week numbering.
ISO3 Tuesday, ISO standard week numbering.
ISO4 Wednesday, ISO standard week numbering.
ISO5 Thursday, ISO standard week numbering.
ISO6 Friday, ISO standard week numbering.
ISO7 Saturday, ISO standard week numbering.
SIMPLE simple week numbering.
STD Sunday, standard week numbering, alternate for STD1.
STD1 Sunday, standard week numbering.
STD2 Monday, standard week numbering.
STD3 Tuesday, standard week numbering.
STD4 Wednesday, standard week numbering.
STD5 Thursday, standard week numbering.
STD6 Friday, standard week numbering.
STD7 Saturday, standard week numbering.
It looks like Focal Point came thru, No Surprise There, and I don't need to open a case. This message has been edited. Last edited by: Doug,
October 31, 2018, 05:27 PM
David Briars
quote:
...? SET WEEKFIRST It shows as "WEEKFIRST 0" However, I ran that same report with the following code prepended to it, and got the same results. ? SET WEEKFIRST -RUN SET WEEKFIRST = 1...
Oh, ok, that is the issue then.
If the value is '0' then the default is used. And the default for '0' is Sunday (as if you had set the value to 1).
Sounds like you have a requirement that Monday is the first day of the week.
If that is the case SET WEEKFIRST to 2 as I show. (Or ISO2 if you would like ISO standard week processing.)
November 01, 2018, 10:06 AM
Doug
Yes Sir, my requirement WAS that Monday is the first day of the week. Now it's Sunday.
BabakNYC ... Me too ... Even with the SET WEEKFIRST ... 
This message has been edited. Last edited by: Doug,