Database

26-Jun-04 9:52

I think I'm either getting over my head, have complicated the issue beyond the scope of SQL, or maybe I have some gaps of knowledge missing for SQL queries. Irregardless, I will try to present the idea, my attempted SQL query, and the issue.

I have 7 tables I want to join (seriously).

Table 1: ERAS - EraCode, Description
Table 2: AGES - EraCode, AgeCode, Description
Table 3: GROUPS - GroupCode, Description
Table 4: CATEGORIES - GroupCode, CategoryCode, Description
Table 5: SKILLS - GroupCode, CategoryCode, SkillCode, Description
Table 6: SUBSKILLS - GroupCode, CategoryCode, SkillCode, SubSkillCode, Description
Table 7: SKILLSINAGES - EraCode, AgeCode, GroupCode, CategoryCode, SkillCode, SubSkillCode

Assume these table have a one to many relationship on the fields that are common.

I created a successful query in Access that manages to get Table 3, 4, 5, and 6 together

<br />
SELECT s.Synopsis, ss.Synopsis, ss.Class, ss.Progression, s.Class, s.Progression, c.Progression, c.GroupCode, c.CategoryCode, s.SkillCode, ss.SubSkillCode, cg.Description, c.Description, s.Description, ss.Description, ss.Table FROM ((CATEGORYGROUPS AS cg LEFT JOIN CATEGORY AS c ON cg.GroupCode = c.GroupCode) LEFT JOIN SKILLS AS s ON (c.CategoryCode = s.CategoryCode) AND (c.GroupCode = s.GroupCode)) LEFT JOIN SUBSKILLS AS ss ON (s.GroupCode = ss.GroupCode) AND (s.CategoryCode = ss.CategoryCode) AND (s.SkillCode = ss.SkillCode) ORDER BY cg.GroupCode, c.CategoryCode, s.SkillCode, ss.SubSkillCode;";<br />

But now I need to add in more restrictions which include Table 1 and 2, therefore I created table 7 as the "relational table" if that is the correct term but I am having difficulty in applying it to my previous query. I though I could do this

<br />
SELECT s.Synopsis, ss.Synopsis, ss.Class, ss.Progression, s.Class, s.Progression, c.Progression, c.GroupCode, c.CategoryCode, s.SkillCode, ss.SubSkillCode, cg.Description, c.Description, s.Description, ss.Description, ss.Table FROM (((CATEGORYGROUPS AS cg LEFT JOIN CATEGORY AS c ON cg.GroupCode = c.GroupCode) LEFT JOIN SKILLS AS s ON (c.CategoryCode = s.CategoryCode) AND (c.GroupCode = s.GroupCode)) LEFT JOIN SUBSKILLS AS ss ON (s.GroupCode = ss.GroupCode) AND (s.CategoryCode = ss.CategoryCode) AND (s.SkillCode = ss.SkillCode)) LEFT JOIN SKILLSINAGES sia ON (cg.GroupCode = sia.GroupCode) AND (c.CategoryCode = sia.CategoryCode) AND (s.SkillCode = sia.SkillCode) AND (ss.SubSkillCode = sia.SubSkillCode) WHERE sia.EraCode LIKE 'ERANAME' AND sia.AgeCode LIKE 'AGENAME' ORDER BY cg.GroupCode, c.CategoryCode, s.SkillCode, ss.SubSkillCode;<br />

But I get an error in access that says the Join operation is not supported? Does anyone know what that means? (a completely different question) Does this design make sense?

Thanks!

Michael Potter28-Jun-04 7:07

28-Jun-04 7:07

Your database design reminds me of old main frame DB designs. They did not have random access to the data (reel to reel tapes) so they duplicated the data across a lot of tables.

I think you have to make better use of Primary Keys and stop the data duplication. First define your descrete entities. Find the attributes that unique to those entities - ignore the attributes that connect entities.

It is hard to give you a sample structure without knowing what the data describes but, things do not look correct.

As an Example: SKILLSINAGES contains an EraCode and an AgeCode. AGES also contains an EraCode. How or why would you keep them in sync? Maybe you need both because they can be different.

I think a better DB design would go a long way in eliminating your query problems.

brdavid28-Jun-04 9:11

28-Jun-04 9:11

the data simply describes skills a person may have where these skills are categorized into Groups, Categories, Skills and SubSkills. Also these skills could be different across time spans, so I have Era and Ages which are categories within an Era. So at the top we have an Era, within that an Age, within that Groups, and so on to Categories, Skills and Subskills.

I think I know what you mean however let me make sure I understand what you are proposing.

I could perhaps make ERAS just a table of Eras, same with AGES, GROUPS, CATEGORIES, SKILLS, and SUBSKILLS. Then define a table that combines the keys for those tables into one table called RELATIONALSKILLS or something. This table would be large, that is it's not improbable for there to be 1000's of records in it.

I think that makes sense, so then, let me ask another question. Say this implementation is correct. Now I make my Query to RELATIONALSKILLS, and I want to gain access to the ERA, AGES, etc tables. Would I still have to perform joins or is there a way to setup the database where I gain access to those fields automagically?

Thanks for the feedback. It may humiliating, but it shows I have a lot more to learn. I would rather make a strong relational table than have a application suffer through a clunky and weak designed Table. Wink | ;)

Michael Potter28-Jun-04 10:37

28-Jun-04 10:37

You have identified the need for a time-span. We can attempt to solve this task first. Ignore the rest of the problem right now.

Questions:
1) How many ERAS are there?
2) How many AGES per ERAS are there?
3) Do AGES overlap ERAS in time?
3) Do AGES descriptions repeat between ERAS?
4) Is this basically a fixed list once it is in production?

You may be over doing the relationship. You may only need one table to represent both ideas if the numbers are small enough and there is no overlap.

CalendarSpan
============
CalendarSpanId    INTEGER     --AutoNumber(Identity) primary Key
TimeOrder         INTEGER     --The order of when this Era/Age occurs (0=Earliest)
Era               TEXT(30)    --The Era 
Age               TEXT(30)    --The Age within the Era

brdavid28-Jun-04 12:04

28-Jun-04 12:04

Well.. to answer

1) unknown, but currently this has only one record. A user has the ability to add more.
2) unknown, but currently there are only 11 ages per the one era.
3) No. Ages are defined to be a unique time span
4) It's possible to have an age that could exist in multiple eras but there wouldn't be a connection there other than the name of the age being the same.
5) No. I'm working on a interface to add, edit and delete.

Also, I defined a field to hold a "Code" version of the Description, that is I stripped all characters but alpha/numeric characters (such as periods, dashes, colons, etc). I did this for searching purposes. In your own personal opinion, is this worth it? or no.

Michael Potter29-Jun-04 8:55

29-Jun-04 8:55

I am assuming the ERACODE and AGECODE are Primary Keys that are never displayed to the user. If this is the case then your time-spans are defined correctly. You are not storing any sequence information but, that might not be important for you application.

Currently, you have 11 AGES records. If the count is not going to be enourmous (100+), I would just load them all into memory and use RegEx to do any fancy searches. I always try to never store calculated data - it can make your edit code more difficult. You can also create a query in Access that handles the calculation on the fly and query against that.

Currently, I would only remove ERACODE from the SKILLSINAGES since AGES already contains a link to it. It is unnecessary duplication. Of course, you will have to back it up with a unique constraint on the EARCODE + AGES.DESCRIPTION Columns. You don't want your users defining two AGES within the same ERAS that have the same DESCRIPTION.

Your next chain looks like this:

GROUPS
|----CATEGORIES
     |----SKILLS
          |----SUBSKILLS

Knowing the SUBSKILLS you should be able to get the SKILLS, CATEGORIES and GROUPS without a problem. There is no need to duplicate this information within each table.

Your SKILLSINAGES table becomes.

SKILLSINAGES
=================
SkillInAgesId - Primary Key
AgeCode
SubSkillCode

Some people would use AgeCode + SubSkillCode as the primary key. I have always liked PK's that have no meaning other than to uniquely identify the row. There are good arguements on both sides of the issue.

brdavid30-Jun-04 2:37

30-Jun-04 2:37

Well, unfortunately, there is more complication to SubSkills than one would think. I have a field in SubSkills called Table that is a Yes/No field. If marked then the SubSkill Code Field (these are just fields that work like SeqNum-autonumbers but they have more meaning than just some arbitrary number) will contain a user-defined table. We can return to this later.

In regards to your recommendations I have adjusted these tables as far as I think they can logically go.

ERAS - EraCode, Description, Synopsis
AGESISERAS - EraCode, AgeCode, Age, TimeOrder, Synopsis
GROUPS - GroupCode, Description
CATEGORIES - GroupCode, CategoryCode, Description, ...
SKILLS - SkillCode, Description, ...
SUBSKILLS - SkillCode, SubSkillCode, Description, ...

the elipse means there are more fields which are dependent on the Description field. I need to have a link to GroupCode from CategoryCode because there might be a Category that appears in two Groups, for example, the Category "General". So I just can't eliminate GroupCode from CATEGORIES since I can't be too sure I will grab the right Group. However, each skill should be unique per Category. I don't see how a person could categorize the same skill in two places. So I have made that definition to eliminate some columns there, however SubSkills needs to know what Skill because there could be a SubSkill that exists in two or more skills, for example, the user defined tables. If I have a Skill called Riding and another called Herding, both of these skills could in turn have the same subskills so I would be unsure what skill I am grabbing if I just looked at SubSkills.

In light of this I think I have gone down as far can, however i light of what you said I think the best I can do for SKILLSINAGES is the following

AgeCode
GroupCode
CategoryCode
SkillCode
SubSkillCode

Because I could have the same Group in different Ages, the same Category in different Groups and the same Subskill in different skills. Maybe there is something I am missing.

thanks for your help in this matter though. It is helping me see my database data a touch differently in light of connections, relations and such. Smile | :)

side question: I could combine tables but wondering if that is also worth it. That is, take for example Groups and Categories. I could combine them so that I have
CATEGORYGROUPS: CategoryGroupID, Group, Category, (other fields here)
I could perform the same with Skills and SubSkills and get
SKILLS: SkillsID, Skill, SubSkill, (other fields here)

A forseeable issue is that not all Skills have SubSkills, so SubSkill may be blank. Does this seem like a logical solution and perhaps a better design? Thanks!

How Can I Join more than two tables in SQL

Grimolfr28-Jun-04 9:58

Grimolfr

28-Jun-04 9:58

I think your problem is all the extra unnecessary parens you have. If you'll notice, each of your original joins only joins with a table one level deeper than the join. (For instance, ss is only joined to s, not to c or cg.)

But when you added the new join, you tried to join all the way back to the deepest level. (i.e. you're trying to join all the way back to cg.)

It's easier to visualize here:

SELECT
		...
FROM
		(
			(
				(
					CATEGORYGROUPS AS cg
					LEFT JOIN CATEGORY AS c
						ON cg.GroupCode = c.GroupCode
				)
				LEFT JOIN SKILLS AS s
					ON (c.CategoryCode = s.CategoryCode)
					AND (c.GroupCode = s.GroupCode)
			)
			LEFT JOIN SUBSKILLS AS ss
				ON (s.GroupCode = ss.GroupCode)
				AND (s.CategoryCode = ss.CategoryCode)
				AND (s.SkillCode = ss.SkillCode)
		)
		LEFT JOIN SKILLSINAGES sia
			ON (cg.GroupCode = sia.GroupCode)
			AND (c.CategoryCode = sia.CategoryCode)
			AND (s.SkillCode = sia.SkillCode)
			AND (ss.SubSkillCode = sia.SubSkillCode)
WHERE
		...

I'm not sure about Access, because I haven't used it for years, but if it were on a SQL Server, I'd remove the parens, like this:

SELECT
		...
FROM
		CATEGORYGROUPS AS cg
		LEFT JOIN CATEGORY AS c
			ON cg.GroupCode = c.GroupCode
		LEFT JOIN SKILLS AS s
			ON (c.CategoryCode = s.CategoryCode)
			AND (c.GroupCode = s.GroupCode)
		LEFT JOIN SUBSKILLS AS ss
			ON (s.GroupCode = ss.GroupCode)
			AND (s.CategoryCode = ss.CategoryCode)
			AND (s.SkillCode = ss.SkillCode)
		LEFT JOIN SKILLSINAGES sia
			ON (cg.GroupCode = sia.GroupCode)
			AND (c.CategoryCode = sia.CategoryCode)
			AND (s.SkillCode = sia.SkillCode)
			AND (ss.SubSkillCode = sia.SubSkillCode)
WHERE
		...

I think the deep nesting is confusing the parser.

Grim

(aka Toby)

MCDBA, MCSD, MCP+SB

Need a Second Life?[^]

SELECT * FROM user WHERE clue IS NOT NULL
GO

(0 row(s) affected)

Inserting Rows...

Sai197125-Jun-04 19:59

Sai1971

25-Jun-04 19:59

Dear All,

In sql-server2000
In a table i am having a column of datatype varchar(8000).
While inserting the record through executenonquery, i am insert only
255 characters rest of the characters are getting trucated.

My question in how i will able to insert the row of that particular
column more than 255 characters

Thanx in advance.

Regards

Re: Inserting Rows...

Rein Hillmann26-Jun-04 22:11

Rein Hillmann

26-Jun-04 22:11

It's not the insert that gets truncated, it's the select from the table.

In Query Analyser set the following option:
Tools->Options->Results->Maximum Characters Per Column=8000

By the way, you can see how much data is in the column by using the datalength() function.

JawedVikia8024-Jun-04 21:40

JawedVikia80

24-Jun-04 21:40

Hi friends
This is my first post in the Forum.
How Can I Join more than two tables in SQL.

Re: How Can I Join more than two tables in SQL

Colin Angus Mackay24-Jun-04 22:03

Colin Angus Mackay

24-Jun-04 22:03

JawedVikia80 wrote:
How Can I Join more than two tables in SQL.

Same way as you join two tables, just keep using joins until you've joined all the tables you need.

For Example:

SELECT TableA.ID as AID, TableB.ID as BID, TableC.ID as CID
FROM TableA
INNER JOIN TableB ON TableA.ID = TableB.AID
INNER JOIN TableC ON TableB.ID = TableC.BID

This example supposes there are one-to-many relationships between TableA & TableB and also between TableB & TableC. And that TableB has a foreign key to tableA called AID, and also TableC has a foreign key to TableB called BID

Does this help?

"You can have everything in life you want if you will just help enough other people get what they want." --Zig Ziglar

The Second EuroCPian Event will be in Brussels on the 4th of September

Can't manage to P/Invoke that Win32 API in .NET? Why not do interop the wiki way!

My Blog

Re: How Can I Join more than two tables in SQL

wgdesigner25-Jun-04 1:38

wgdesigner

25-Jun-04 1:38

Dear Friend There are many types of the Joins. which type of the join u want to use??

1-Inner join
2-Outer join
3-Self Join

You have example of inner join by our one friend if u need other 2's then reply..

Raj Khatri

Concurrency

IamADotNetGuy24-Jun-04 10:04

IamADotNetGuy

24-Jun-04 10:04

Hi
Please dont get upset if u see this message in two places,
As it is bit urgent i have posted at two place one is at c# forums and another is at this place

How can i handle concurrency in DotNet?
I need to update a record into a database table, by the time I will access the record there is a chance that somebody can change the record.

I know this kind of problem is common.

But never i had an oppurtunity to work this kind of problem.

please let me know the solution.

Re: Concurrency

Rein Hillmann26-Jun-04 22:13

Rein Hillmann

26-Jun-04 22:13

Do a search in SQL Books Online for "transactions". More specifically:

begin transaction
commit transaction
and
rollback transaction

Re: Concurrency

Grimolfr28-Jun-04 7:56

Grimolfr

28-Jun-04 7:56

Also, assuming that you're using SQL Server 2000 (or MSDE) have a look at:

sp_getapplock
sp_releaseapplock

In most of the situations where our OLTP system has concurrency issues, I've found app locks to be infinitely more useful than transactions. Transactions (IMHO) should be used for atomicity, not for concurrency issues.

Grim

(aka Toby)

MCDBA, MCSD, MCP+SB

Need a Second Life?[^]

Re: Concurrency

IamADotNetGuy28-Jun-04 8:07

IamADotNetGuy

28-Jun-04 8:07

This will work.....
But i am using Oracle and also i want to handle this at BLL but not in database.

SQL Query

Guinness4Strength24-Jun-04 9:39

Guinness4Strength

24-Jun-04 9:39

Couple of Query questions:
1. is there a way to step through a query and view variables in the Anaylzer for MS SQL Server ?
2. I'm attempting to create a query that will be run dailly for the previous day's infomation, I'm having trouble getting the previous date in a usable format.
The following uis what i have so far:

DECLARE @Year int
DECLARE @Month int
DECLARE @Day int
DECLARE @FullDate varchar(10)
DECLARE @Yesterday datetime
SET @Yesterday=DATEADD(day,-1,GETDATE())
SET @Year = CAST(DATEPART(yyyy,@Yesterday) as varchar(4))
SET @Month = CAST(DATEPART(mm,@Yesterday) as varchar(2))
SET @Day = CAST(DATEPART(dd,@Yesterday) as varchar(2))
SET @FullDate=@Year+'-'+@Month+'-'+@Day
SELECT * FROM Document WHERE (ProcessDate=@FullDate)

Can someone point out where I'm going wrong ?

Re: SQL Query

Michael Potter25-Jun-04 3:32