[10g] Best way to populate and update multiple workday calendars?

Question

I would like to create multiple workday calendars, and be able to update them as needed. Each workday calendar would be for at most 5 years (right now), although it is possible that at some point in the future, I might want to extend that.
A workday calendar can apply to a single resource or a group of resources. Each individual is part of a resource group. What I'm thinking is that if an individual resource has an associated workday calendar, that calendar gets used, but if it doesn't, it uses the calendar of it's resource group, and if the group doesn't have one, the default calendar gets used. Theoretically, each resource could have its own workday calendar, and there would be probably 500 resources at most. In practice, many resources will have the same or similar workday calendars.
Each calendar will be based on one of 3 standards: all calendars days are work days, all days but weekends are workdays, or all days except weekends and holidays are workdays. (Weeks start on Sunday, and Saturday and Sunday are weekends.) The standard calendar would then be modified to create each unique calendar as needed. For example, if a resource was an employee, their calendar might be the standard of not working weekends or holidays, but could also include a one week vacation in February and a one week vacation July. I'm not sure what the best approach is for defining a calendar in the first place and then being able to update it as an employee decides on vacations (or whatever other situation might affect workdays).
Furthermore, I'd really like to be able to incorporate work hours, which could vary by day, though would likely be fairly standard. I don't know if this information belongs in the workday table, or as something separate to be combined with the workday table.
My ultimate purpose in doing all of this is to try to schedule a very large project amongst many resources.
Here is some sample data illustrating where I'm at so far:
CREATE TABLE     work_groups
(     group_id     VARCHAR2(5)     NOT NULL
,     group_name     VARCHAR2(25)     
,     group_desc     VARCHAR2(200)
,     CONSTRAINT     work_groups_pk     PRIMARY KEY (group_id)
INSERT INTO     work_groups
VALUES     ('A','Group A','Group A description');
INSERT INTO     work_groups
VALUES     ('B','Group B','Group B description');
INSERT INTO     work_groups
VALUES     ('C','Group C','Group C description');
INSERT INTO     work_groups
VALUES     ('D','Group D','Group D description');
CREATE TABLE     resources
(     resource_id     VARCHAR2(20)     NOT NULL
,     type          VARCHAR2(1)
,     description     VARCHAR2(200)
,     group_id     VARCHAR2(5)     
,     CONSTRAINT     resources_pk     PRIMARY KEY (resource_id)
,     CONSTRAINT     group_id_fk     FOREIGN KEY (group_id)
                         REFERENCES  work_groups (group_id)
INSERT INTO     resources
VALUES     ('A001','M','text here','A');
INSERT INTO     resources
VALUES     ('A002','M','text here','A');
INSERT INTO     resources
VALUES     ('A003','M','text here','A');
INSERT INTO     resources
VALUES     ('B001','M','text here','B');
INSERT INTO     resources
VALUES     ('B002','M','text here','B');
INSERT INTO     resources
VALUES     ('C001','M','text here','C');
INSERT INTO     resources
VALUES     ('C002','M','text here','C');
INSERT INTO     resources
VALUES     ('C003','M','text here','C');
INSERT INTO     resources
VALUES     ('D001','M','text here','D');
INSERT INTO     resources
VALUES     ('12345','L','text here','A');
INSERT INTO     resources
VALUES     ('12346','L','text here','A');
INSERT INTO     resources
VALUES     ('12347','L','text here','B');
INSERT INTO     resources
VALUES     ('12348','L','text here','B');
INSERT INTO     resources
VALUES     ('12349','L','text here','C');
INSERT INTO     resources
VALUES     ('12350','L','text here','C');
INSERT INTO     resources
VALUES     ('12351','L','text here','D');I'm not sure if I should have a separate table to define a relationship between a resource or resource group and a calendar id (each resource or group would only be able to be assigned 1 unique calendar id, though multiple resources/groups could share the same calendar id), or if I should add an extra column to each of the above tables to assign the calendar id.
CREATE TABLE     calendars
(     cal_id          NUMBER(4)     NOT NULL
,     cal_title     VARCHAR2(25)
,     cal_desc     VARCHAR2(200)
,     CONSTRAINT     calendars_pk     PRIMARY KEY (cal_id)
INSERT INTO     calendars
VALUES     (1,'Default','This is the default calendar to use for workdays');
INSERT INTO     calendars
VALUES     (2,'All Days','This calendar treats all days as workdays');
INSERT INTO     calendars
VALUES     (3,'Weekends Off','This calendar gives weekends off, but no holidays');
INSERT INTO     calendars
VALUES     (4,'Holidays Off','This calendar gives weekends and holidays off');
CREATE TABLE     workdays
(     cal_id          NUMBER(4)     NOT NULL
,     cal_date     DATE          NOT NULL
,     cal_year     NUMBER(4)
,     work_day     NUMBER(3)
,     work_date     DATE
,     work_week     NUMBER(2)
,     work_year     NUMBER(4)
,     work_days     NUMBER(5)
,     cal_days     NUMBER(6)
,     CONSTRAINT     workdays_pk     PRIMARY KEY (cal_id, cal_date)
,     CONSTRAINT     cal_id_fk     FOREIGN KEY (cal_id)
                         REFERENCES  calendars (cal_id)
);cal_id     - refers to calendars table
cal_date - the actual calendar date
cal_year - the actual calendar year for the given calendar date
work_day - workday in that work year (resets each year, starting at 1, is 0 if that calendar date is not a work day)
work_date - if a work day, the calendar date, if not, the calendar date of the last work day (or in the first week of the calendar, the next work day)
work_week - work week of the work date (numbered starting at 1 , reset each year at the first Sunday of the year, before first Sunday will have last week of previous year, and the first year of the calendar will have all days prior to Sunday included in the first week, so the first week of a calendar could be more than 7 days)
work_year - year of the work date
work_days - shop day of the work date (except in first week of the calendar, before first shop day is 0), starts at 1 (at the beginning of the calendar), cummulative (does not reset each year)
cal_days - calendar day of the work date, starts at 1 (at the beginning of the calendar), cummulative (does not reset each year)
Assuming the calendar starts on 1/1/2010 (these values are approximately correct--I just made my best guess to provide sample data):
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/01/2010','mm/dd/yyyy'),2010,0,TO_DATE('01/04/2010','mm/dd/yyyy'),1,2010,0,1);
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/02/2010','mm/dd/yyyy'),2010,0,TO_DATE('01/04/2010','mm/dd/yyyy'),1,2010,0,2);
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/03/2010','mm/dd/yyyy'),2010,0,TO_DATE('01/04/2010','mm/dd/yyyy'),1,2010,0,3);
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/04/2010','mm/dd/yyyy'),2010,1,TO_DATE('01/04/2010','mm/dd/yyyy'),1,2010,1,4);
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/05/2010','mm/dd/yyyy'),2010,2,TO_DATE('01/05/2010','mm/dd/yyyy'),1,2010,2,5);
INSERT INTO     workdays
VALUES     (3, TO_DATE('12/23/2010','mm/dd/yyyy'),2010,250,TO_DATE('12/23/2010','mm/dd/yyyy'),51,2010,250,357);
INSERT INTO     workdays
VALUES     (3, TO_DATE('12/24/2010','mm/dd/yyyy'),2010,0,TO_DATE('12/23/2010','mm/dd/yyyy'),51,2010,250,358);
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/01/2011','mm/dd/yyyy'),2011,0,TO_DATE('12/23/2010','mm/dd/yyyy'),51,2010,250,366);
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/02/2011','mm/dd/yyyy'),2011,0,TO_DATE('12/23/2010','mm/dd/yyyy'),1,2011,250,367);
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/03/2011','mm/dd/yyyy'),2011,1,TO_DATE('01/03/2010','mm/dd/yyyy'),1,2011,251,368);I've tried googling workday calendars and similar things, but I can't seem to find anything that fits my situation. If anyone could even point me in the right direction, I'd appreciate it.
I'm working in 10g (XE).
Edited by: user11033437 on Jul 19, 2011 3:05 PM
Also, I don't know if it would make more sense to just somehow store the days each resource/group doesn't work, and generate a workday calendar on the fly as needed, rather than trying to store possibly thousands of dates in the database??

Answer 1

Hi,
Interesting problem!
I'm not sure exactly what you want, however. Are you looking for a way to answer questions such as "For resource A001, what are the first 6 work days on or after January 4, 2010?" or "How many work days does resource A001 have between January 4 and January 12, 2010?" Post a few examples of the questions you might ask, and the results you would want, given the sample data you posted.
user11033437 wrote:
I would like to create multiple workday calendars, and be able to update them as needed. Each workday calendar would be for at most 5 years (right now), although it is possible that at some point in the future, I might want to extend that.
A workday calendar can apply to a single resource or a group of resources. Each individual is part of a resource group. Is a "resouce group" the same as a "work group"?
if a resource moves from one resource group to another, do you need to keep track of the historical information? For example, if resource A001 does not havfe its own calendar, and is part of work_group A on Juanuary 1, 2010, but then moves to work_group B on July 1, 2010, will you need to answer questions like "How many work days did A001 have in 2010", where you have to remember that the work_group A calendar apllied during the first half of the year, but the work_group B calendar was in use for the second half?
What I'm thinking is that if an individual resource has an associated workday calendar, that calendar gets used, but if it doesn't, it uses the calendar of it's resource group, and if the group doesn't have one, the default calendar gets used. Theoretically, each resource could have its own workday calendar, and there would be probably 500 resources at most. In practice, many resources will have the same or similar workday calendars.
Each calendar will be based on one of 3 standards: all calendars days are work days, all days but weekends are workdays, or all days except weekends and holidays are workdays. (Weeks start on Sunday, and Saturday and Sunday are weekends.) The standard calendar would then be modified to create each unique calendar as needed. For example, if a resource was an employee, their calendar might be the standard of not working weekends or holidays, but could also include a one week vacation in February and a one week vacation July. I'm not sure what the best approach is for defining a calendar in the first place and then being able to update it as an employee decides on vacations (or whatever other situation might affect workdays).It seems like the simplest thing would be to record only the exceptions to the base calendar. That is, given that the employee normally abides by the "no weekends or holidays" schedule, just enter 5 rows for that particular employee to mark the 5 work days he'll miss in February. If the emplyoee is going to work Saturdays in June (in addition to his regular schedule), then enter one row for each Saturday in June.
>
Furthermore, I'd really like to be able to incorporate work hours, which could vary by day, though would likely be fairly standard. I don't know if this information belongs in the workday table, or as something separate to be combined with the workday table. That depends on exactly what you want. Post a couple opf examples of questions you would want to answer, and the actual answers, given the sample data you post.
My ultimate purpose in doing all of this is to try to schedule a very large project amongst many resources.
Here is some sample data illustrating where I'm at so far:
CREATE TABLE     work_groups
(     group_id     VARCHAR2(5)     NOT NULL
,     group_name     VARCHAR2(25)     
,     group_desc     VARCHAR2(200)
,     CONSTRAINT     work_groups_pk     PRIMARY KEY (group_id)
INSERT INTO     work_groups
VALUES     ('A','Group A','Group A description');
INSERT INTO     work_groups
VALUES     ('B','Group B','Group B description');
INSERT INTO     work_groups
VALUES     ('C','Group C','Group C description');
INSERT INTO     work_groups
VALUES     ('D','Group D','Group D description');
CREATE TABLE     resources
(     resource_id     VARCHAR2(20)     NOT NULL
,     type          VARCHAR2(1)
,     description     VARCHAR2(200)
,     group_id     VARCHAR2(5)     
,     CONSTRAINT     resources_pk     PRIMARY KEY (resource_id)
,     CONSTRAINT     group_id_fk     FOREIGN KEY (group_id)
                         REFERENCES  work_groups (group_id)
INSERT INTO     resources
VALUES     ('A001','M','text here','A');
INSERT INTO     resources
VALUES     ('A002','M','text here','A');
INSERT INTO     resources
VALUES     ('A003','M','text here','A');
INSERT INTO     resources
VALUES     ('B001','M','text here','B');
INSERT INTO     resources
VALUES     ('B002','M','text here','B');
INSERT INTO     resources
VALUES     ('C001','M','text here','C');
INSERT INTO     resources
VALUES     ('C002','M','text here','C');
INSERT INTO     resources
VALUES     ('C003','M','text here','C');
INSERT INTO     resources
VALUES     ('D001','M','text here','D');
INSERT INTO     resources
VALUES     ('12345','L','text here','A');
INSERT INTO     resources
VALUES     ('12346','L','text here','A');
INSERT INTO     resources
VALUES     ('12347','L','text here','B');
INSERT INTO     resources
VALUES     ('12348','L','text here','B');
INSERT INTO     resources
VALUES     ('12349','L','text here','C');
INSERT INTO     resources
VALUES     ('12350','L','text here','C');
INSERT INTO     resources
VALUES     ('12351','L','text here','D');
It looks like all the rows have the same description. if description matters in this problem, wouldn't it be better to illustrate how it matters, by having different descrioptions that appeared in different outputs? On the other hand, if description doesn't play any role in this problem, then why include in the sample data at all?
I'm not sure if I should have a separate table to define a relationship between a resource or resource group and a calendar id (each resource or group would only be able to be assigned 1 unique calendar id, though multiple resources/groups could share the same calendar id), or if I should add an extra column to each of the above tables to assign the calendar id.
CREATE TABLE     calendars
(     cal_id          NUMBER(4)     NOT NULL
,     cal_title     VARCHAR2(25)
,     cal_desc     VARCHAR2(200)
,     CONSTRAINT     calendars_pk     PRIMARY KEY (cal_id)
INSERT INTO     calendars
VALUES     (1,'Default','This is the default calendar to use for workdays');
INSERT INTO     calendars
VALUES     (2,'All Days','This calendar treats all days as workdays');
INSERT INTO     calendars
VALUES     (3,'Weekends Off','This calendar gives weekends off, but no holidays');
INSERT INTO     calendars
VALUES     (4,'Holidays Off','This calendar gives weekends and holidays off');
What is is cal_id=1? How will it differ from the other three?
CREATE TABLE     workdays
(     cal_id          NUMBER(4)     NOT NULL
,     cal_date     DATE          NOT NULL
,     cal_year     NUMBER(4)
,     work_day     NUMBER(3)
,     work_date     DATE
,     work_week     NUMBER(2)
,     work_year     NUMBER(4)
,     work_days     NUMBER(5)
,     cal_days     NUMBER(6)
,     CONSTRAINT     workdays_pk     PRIMARY KEY (cal_id, cal_date)
,     CONSTRAINT     cal_id_fk     FOREIGN KEY (cal_id)
                         REFERENCES  calendars (cal_id)
);I suspect there's a simpler way, especially if there's a regular order to day types (e.g., people who get holidays off normally get weekeneds, too).
Perhaps you could have a table like this, that had one row per day:
CREATE TABLE  days
(       a_date      DATE     PRIMARY KEY
,     day_type    NUMBER (1)              -- 1=Weekend, 2=Holiday, 3=Other
INSERT INTO days (a_date, day_type) VALUES (DATE '2010-01-01', 2) /* New Years Day */;
INSERT INTO days (a_date, day_type) VALUES (DATE '2010-01-02', 1) /* Saturday */;
INSERT INTO days (a_date, day_type) VALUES (DATE '2010-01-03', 1) /* Sunday */;
INSERT INTO days (a_date, day_type) VALUES (DATE '2010-01-04', 3) /* Monday - back to work */;
...Another table (I'll call it work_sched) shows what resources work when:
CREATE TABLE  work_sched
(       p_key          NUMEBR     PRIMARY KEY     -- Arbitrary Unique ID
,     group_id         VARCHAR2 (5)          -- Exactly one of the columns group_id or ...
,     resource_id      VARCHAR2 (20)          --     ... resource_id will always be NULL
,     a_date              DATE
,     works_on     NUMBER (1)          -- works when days.day_type >= this value
,     remarks          VARCHAR2 (40)
);To indicate that work_group 'L' normally works on type 3 days only (that is, weekends or holidays):
INSERT INTO work_sched (group_id, a_date, works_on) VALUES ('L', NULL, 3);(Let's assume that p_key is populated by a trigger.)
The NULL in the a_date column indicates that this applies to all days, unless overridden by another row in the work_sched table. Instead of NULL, mabe some impossible date (such as Jan. 1, 1900) would be more convenient to indicate defaults.
Exceptions to this schedule would be indicated by other rows in work_sched. For example, if '12345' is the employee who's on vacation for a week in February:
INSERT INTO work_sched (resource_id, a_date, works_on, remarks) VALUES ('12345', DATE '2010-02-08', 4, 'Vacation');
INSERT INTO work_sched (resource_id, a_date, works_on, remarks) VALUES ('12345', DATE '2010-02-09', 4, 'Vacation');
INSERT INTO work_sched (resource_id, a_date, works_on, remarks) VALUES ('12345', DATE '2010-02-10', 4, 'Vacation');
INSERT INTO work_sched (resource_id, a_date, works_on, remarks) VALUES ('12345', DATE '2010-02-11', 4, 'Vacation');
INSERT INTO work_sched (resource_id, a_date, works_on, remarks) VALUES ('12345', DATE '2010-02-12', 4, 'Vacation');And if that same employee works Satudays in June:
INSERT INTO work_sched (resource_id, a_date, works_on, remarks) VALUES ('12345', DATE '2010-06-05', 1, 'Fiscal year-end crunch');
INSERT INTO work_sched (resource_id, a_date, works_on, remarks) VALUES ('12345', DATE '2010-06-12', 1, 'Fiscal year-end crunch');
INSERT INTO work_sched (resource_id, a_date, works_on, remarks) VALUES ('12345', DATE '2010-06-19', 1, 'Fiscal year-end crunch');
INSERT INTO work_sched (resource_id, a_date, works_on, remarks) VALUES ('12345', DATE '2010-06-26', 1, 'Fiscal year-end crunch');When finding the number of work days, we would join work_sched to days using both conditions:
ON   work_sched.date           = days.a_date
AND  work_sched.works_on  <= days.day_type
cal_id     - refers to calendars table
cal_date - the actual calendar date
cal_year - the actual calendar year for the given calendar date
work_day - workday in that work year (resets each year, starting at 1, is 0 if that calendar date is not a work day)
work_date - if a work day, the calendar date, if not, the calendar date of the last work day (or in the first week of the calendar, the next work day)
work_week - work week of the work date (numbered starting at 1 , reset each year at the first Sunday of the year, before first Sunday will have last week of previous year, and the first year of the calendar will have all days prior to Sunday included in the first week, so the first week of a calendar could be more than 7 days)
work_year - year of the work date
work_days - shop day of the work date (except in first week of the calendar, before first shop day is 0), starts at 1 (at the beginning of the calendar), cummulative (does not reset each year)
cal_days - calendar day of the work date, starts at 1 (at the beginning of the calendar), cummulative (does not reset each year)This is a lot of denormalized data. that is, you should be able to easily deduce cal_year from cal_date, but sometimes it is convenient to store denormalized data.
Assuming the calendar starts on 1/1/2010 (these values are approximately correct--I just made my best guess to provide sample data):
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/01/2010','mm/dd/yyyy'),2010,0,TO_DATE('01/04/2010','mm/dd/yyyy'),1,2010,0,1);
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/02/2010','mm/dd/yyyy'),2010,0,TO_DATE('01/04/2010','mm/dd/yyyy'),1,2010,0,2);
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/03/2010','mm/dd/yyyy'),2010,0,TO_DATE('01/04/2010','mm/dd/yyyy'),1,2010,0,3);
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/04/2010','mm/dd/yyyy'),2010,1,TO_DATE('01/04/2010','mm/dd/yyyy'),1,2010,1,4);
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/05/2010','mm/dd/yyyy'),2010,2,TO_DATE('01/05/2010','mm/dd/yyyy'),1,2010,2,5);
INSERT INTO     workdays
VALUES     (3, TO_DATE('12/23/2010','mm/dd/yyyy'),2010,250,TO_DATE('12/23/2010','mm/dd/yyyy'),51,2010,250,357);
INSERT INTO     workdays
VALUES     (3, TO_DATE('12/24/2010','mm/dd/yyyy'),2010,0,TO_DATE('12/23/2010','mm/dd/yyyy'),51,2010,250,358);
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/01/2011','mm/dd/yyyy'),2011,0,TO_DATE('12/23/2010','mm/dd/yyyy'),51,2010,250,366);
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/02/2011','mm/dd/yyyy'),2011,0,TO_DATE('12/23/2010','mm/dd/yyyy'),1,2011,250,367);
INSERT INTO     workdays
VALUES     (3, TO_DATE('01/03/2011','mm/dd/yyyy'),2011,1,TO_DATE('01/03/2010','mm/dd/yyyy'),1,2011,251,368);I've tried googling workday calendars and similar things, but I can't seem to find anything that fits my situation. If anyone could even point me in the right direction, I'd appreciate it.
I'm working in 10g (XE).
Edited by: user11033437 on Jul 19, 2011 3:05 PM
Also, I don't know if it would make more sense to just somehow store the days each resource/group doesn't work, and generate a workday calendar on the fly as needed, rather than trying to store possibly thousands of dates in the database??That's what I was thinking, too.
Post some sample data (if it's not what you've already posted), a couple of sample questions, and the exact answers you want from each question given that sample data.