i'm refactoring a db structure and have a little problem.
This DB have various tables with same structure, like:
People -> People_contacts
Activities -> Activities_contacts
Now, i want to create only one Contact table, and use an ENUM() to distinguish from the nature of the parent (for search requirements and data reversibility)
the structure will be:
People -> Contacts[People]
Activities -> Contacts[Activities]
But now i need to put a Foreign-key, and based on the ENUM property distinguish from two different tables...
How i can effort this? There are a way or is better maintain the old tables?
why you are using view? if the People_contacts and Activities_contacts are exactly the same, you can try this:
create view `test` as select *,'People' as Type from `People_contacts` union select *,'Activities' from `Activities _contacts` union;
and then select what you want from the view:
select * from `test` where Type = 'People' and .....
and your query answer should be this
+----+------+ +--------+
| ID | Data |...| Type |
+----+------+ +--------+
| 1 | foo |...| People |
| 2 | foo |...| People |
+----+------+ +--------+
You cannot have a declared foreign key, pointing to one table or another depending on a field.
You can do a few things, but none of then are really clean.
You can have the integer field, and the enum, but do not declare the field as a foreign key. You will have to implement all the logic by yourself, and it will be harder to maintain, and harder to decouple database from programing.
You can have 2 nullable foreign keys (people_id and activity_id), and forget the enum field. if one FK is null, the other will have the real relation. This is better since you declare the foreign keys as usual and the model is stronger
If you prefer to keep your contact table clean, you can have a relation table where you put this dirty stuff. So in this table you store the contact_id, and the id of the activity or the person, as explained in whatever 1 or 2
But anyways, probably you are obfuscated and you dont need to have the foreign key in the contact table. I would bet you will always access first the people or the activities table, so you probably can change this tables, and add a contact_id foreign key. In the contact table you just need to add, if you dont have it already, de id primary key, and delete de ENUM field, and the foreign keys, since you dont really need them
Related
I have an extreamly simple idea: table that keeps user "achievements". And it is as simple as that:
user_id | achievement_id
1 | 1
1 | 2
1 | 5
2 | 2
2 | 3
All what I need is user id, and id of achievement if he already got it. All what I need to SELECT is SELECT achievement_id WHERE user_id=x. So no need for an artificial autoincrement column that I'll never use or know what it contains. But setting an primary key is required, so the question is - is it good idea to make such 2-column table and set both columns as multi-column primary key? I already have a set of 3-columns table where 2 are primary key, because it is logic... Well, logic for me, but for the database?
These types of tables are common in cases of n-n relationships, multivalued attributes, and weak entities. It varies a lot from its modeling, but yes, it is a good solution for some cases. the primary key is usually the relation of the columns. In your case it would be user_id and achievement_id.
Yes since the rule for such a set of n-keys is: "I only want one kind of record which has this set (a,b) of keys".
-> therefore you won't be able to add twice "Mario, achievement1".
Primary key will be then (PlayerID, AchievementID).
If you want to add some informations about this achievement (for example, when the player got the achievement), simply do such as: (PlayerID, AchievementID, Date) with PlayerID, AchievementID as primary key.
I hope this will help you.
This should be a simple question I think, but is it OK to have NULL foreign keys?
I guess to elaborate, let's say I'm making a database for users and different types of users require different data sets... what would be the best practice to design this database?
This was my thought, as a rough example: (am i correct or way off?)
"users":
id | type (ie. '1' for basic, '2' for advanced) | basic_id (nullable foreign key) | advanced_id (nullable foreign key) | email | name | address | phone (etc etc)
"users_basic":
id | user_id (foreign key) | (other data only required for basic users)
"users_advanced"
id | user_id (forgein key) | (other data only required for advanced users)
I get the feeling it's bad design cause there's no way to get all the data in one query without checking what type of user it is first, but I really don't like the idea of having ONE table with a ton of NULL data. What is the best way to design this?
Of course it is fine to have NULL foreign keys.
In your case, though, I'd be inclined to do one of two things. If there really aren't very many columns for the basic and advanced users, you can just include them in the users table. This would be the typical approach.
Otherwise, you can declare user_id as the primary key in all three tables, and still have a foreign key relationship from the secondary tables (users_basic and users_advanced) to the primary (users). Maintaining the distinctiveness of the relationship is tricky in MySQL and probably not worth doing.
I've created a database with three tables in it:
Restaurant
restaurant_id (autoincrement, PK)
Owner
owner_id (autoincrement, PK)
restaurant_id (FK to Restaurant)
Deal
deal_id (autoincrement)
owner_id (FK to Owner)
restaurant_id (FK to Restaurant)
(PK: deal_id, owner_id, restaurant_id)
There can be many owners for each restaurant. I chose two foreign keys for Deal so I can reference the deal by either the owner or the restaurant. The deal table would have three primary keys, two being foreign keys. And it would have two one-to-many relationships pointing to it. All of my foreign keys are primary keys and I don't know if I'll regret doing it like this later on down the road. Does this design make sense, and seem good for what I'm trying to achieve?
Edit: What I really need to be able to accomplish here is when a owner is logged in and viewing their account, I want them to be able to see and edit all the deals that are associated with that particular restaurant. And because there can be more that one owner per restaurant, I need to be able to perform a query something like: select *from deals where restaurant_id = restaurant_id. In other words, if I'm an owner and I'm logged in, I need to be able to make query: get all of the deal that are related to not just me, the owner, but to all of the owners associated with this restaurant.
You're having some trouble with terminology.
A table can only ever have a one primary key. It is not possible to create a table with two different primary keys. You can create a table with two different unique indexes (which are much like a primary key) but only one primary key can exist.
What you're asking about is whether you should have a composite or compound primary key; a primary key using more than one column.
Your design is okay, but as written you probably have no need for the column deal_id. It seems to me that restaurant_id and owner_id together are enough to uniquely identify a row in Deal. (This may not be true if one owner can have two different ownership stakes in a single restaurant as the result of recapitalization or buying out another owner, but you don't mention anything like that in your problem statement).
In this case, deal_id is largely wasted storage. There might be an argument to be made for using the deal_id column if you have many tables that have foreign keys pointing to Deal, or if you have instances in which you want to display to the user Deals for multiple restaurants and owners at the same time.
If one of those arguments sways you to adopt the deal_id column, then it, and only it, should be the primary key. There would be nothing added by including the other two columns since the autoincrement value itself would be unique.
If u have a unique field, this should be the PK, that would be the incremented field.
In this specific case it gives u nothing at all to add more fields to this key, it actually somewhat impacts performance (don't ask me how much, u bench it).
if you want to create 2 foreign keys in the deal table which are the restaurant and the owner the logic is something like a table could exist in the deal even without an owner or an owner could exist in the deal even without identifying the table on it but you could still identify the table because it's being used as a foreign key on the owner table, but if your going to put values on each columns that you defined as foreign key then I think it's going to be redundant cause I'm not sure how you would use the deal table later on but by it's name I think it speaks like it would be used to identify if a restaurant table is being reserved or not by a customer and to see how you have designed your database you could already identify the table which they have reserved even without specifying the table as foreign key in the deal table cause by the use of the owner table you would able to identify which table they have reserved already since you use it as foreign key on the owner table you just really have to be wise on defining relationships between your tables and avoid redundancy as much as possible. :)
I think it is not best.
First of all, the Deal table PK should be the deal_id. There is no reason to add additional columns to it--and if you did want to refer to the deal_id in another table, you'd have to include the restaurant_id and owner_id which is not good. Whether deal_id should also be the clustered index (a.k.a. index organized on this column) depends on the data access pattern. Will your database be full of data_id values most often used for lookup, or will you primarily be looking deals up by owner_id or restaurant_id?
Also, using two separate FKs way the you have described it (as far as I can tell!) would allow a deal to have an owner and restaurant combination that are not a valid (combining an owner that does not belong to that restaurant). In the Deal table, instead of one FK to Owner and one FK to Restaurant, if you must have both columns, there should be a composite FK to only the Owner table on (OwnerID, RestaurantID) with a corresponding unique key in the Owner table to allow this link up.
However, with such a simple table structure I don't really see the problem in leaving RestaurantID out of the Deal table, since the OwnerID always fully implies the RestaurantID. Obviously your deals cannot be linked only with the restaurant, because that would imply a 1:M relationship on Deal:Owner. The cost of searching based on Restaurant through the Owner table shouldn't really be that bad.
Its not wrong, it works. But, its not recommended.
Autoincrement Primary Keys works without Foreign Keys (or Master Keys)
In some databases, you cannot use several fields as a single primary key.
Compound Primary Keys or Compose Primary Keys are more difficult to handle in a query.
Compound Primary Key Query Example:
SELECT
D.*
FROM
Restaurant AS R,
Owner AS O,
Deal AS D
WHERE
(1=1) AND
(D.RestaurantKey = D.RestaurantKey) AND
(D.OwnerKey = D.OwnerKey)
Versus
Single Primary Key Query Example:
SELECT
D.*
FROM
Restaurant AS R,
Owner AS O,
Deal AS D
WHERE
(D.OwnerKey = O.OwnerKey)
Sometimes, you have to change the value of foreign key of a record, to another record. For Example, your customers already order, the deal record is registered, and they decide to change from one restaurant table to another. So, the data must be updated, in the "Owner", and "Deal" tables.
+-----------+-------------+
| OwnerKey | OwnerName |
+-----------+-------------+
| 1 | Anne Smith |
+-----------+-------------+
| 2 | John Connor |
+-----------+-------------+
| 3 | Mike Doe |
+-----------+-------------+
+-----------+-------------+-------------+
| OwnerKey | DealKey | Food |
+-----------+-------------+-------------+
| 1 | 1 | Hamburguer |
+-----------+-------------+-------------+
| 2 | 2 | Hot-Dog |
+-----------+-------------+-------------+
| 3 | 3 | Hamburguer |
+-----------+-------------+-------------+
| 1 | 3 | Soda |
+-----------+-------------+-------------+
| 2 | 1 | Apple Pie |
+-----------+-------------+-------------+
| 3 | 3 | Chips |
+-----------+-------------+-------------+
If you use compound primary keys, you have to create a new record for "Owner", and new records for "Deals", copy the other fields, and delete the previous records.
If you use single keys, you just have to change the foreign key of Table, without inserting or deleting new records.
Cheers.
Firstly, I apologise if this is a dupe - I suspect it may be but I can't find it.
Say I have a table of companies:
id | company_name
----+--------------
1 | Someone
2 | Someone else
...and a table of contacts:
id | company_id | contact_name | is_primary
----+------------+--------------+------------
1 | 1 | Tom | 1
2 | 2 | Dick | 1
3 | 1 | Harry | 0
4 | 1 | Bob | 0
Is it possible to set up the contacts table in such a way that it requires that one and only one record has the is_primary flag set for each common company_id?
So if I tried to do:
UPDATE contacts
SET is_primary = 1
WHERE id = 4
...the query would fail, because Tom (id = 1) is already flagged as the primary contact for company_id = 1. Or even better, would it be possible to construct a trigger so that the query would succeed, but Tom's is_primary flag would be cleared by the same operation?
I am not too bothered about checking whether company_id exists in the companies table, my PHP code would already have performed this check before I got to this stage (although if there is a way to do this in the same operation it would be nice, I suppose).
When I initially thought about this I thought "that will be easy, I'll just add a unique index across the company_id and is_primary columns" but obviously that won't work as it would restrict me to one primary and one non-primary contact - any attempt to add a third contact would fail. But I can't help feeling there would be a way to configure a unique index that gives me the minimum functionality I require - to reject an attempt to add a second primary contact, or reject an attempt to leave a company with no primary contact.
I am aware that I could just add a primary_contact field to the companies table with an FK to the contacts table but it feels messy. I don't like the idea of both tables having an FK to the other - it seems to me that the one table should rely on the other, not both tables relying on each other. I guess I just think that over time there is more chance of something going wrong.
To sum up:
How can I restrict the contacts table so that one and only one record with a given company_id has the is_primary flag set?
Anyone have any thoughts on whether two tables having FKs to each other is a good/bad idea?
Circular refenences between tables are indeed messy. See this (decade old) article: SQL By Design: The Circular Reference
The cleanest way to make such a constraint is to add another table:
Company_PrimaryContact
----------------------
company_id
contact_id
PRIMARY KEY (company_id)
FOREIGN KEY (company_id, contact_id)
REFERENCES Contact (company_id, id)
This will also require a UNIQUE constraint in table Contact on (company_id, id)
You could just do a query before that one setting
UPDATE contacts SET is_primary = 0 WHERE company_id = .....
or even
UPDATE contacts
SET is_primary = IF(id=[USERID],1,0)
WHERE company_id = (
SELECT company_id FROM contacts WHERE id = [USERID]
);
Just putting an alternative out there - personally I'd probably look to the FK approach though instead of this type of workaround i.e. have a field in the companies table with a primary_user_id field.
EDIT method w/o relying on a contact.is_primary field
Alternative method, first of all remove is_primary from contacts. Secondly add a "primary_contact_id" INT field into companies. Thirdly, when changing the primary contact, just change that primary_contact_id thus preventing any possibility of there being more than 1 primary contact at any time and all without the need for triggers etc in the background.
This option would work fine in any engine as it's simply updating an INT field, any reliance on FK's etc could be added/removed as required but at it's simplest it's just changing an INT fields value
This option is viable as long as you need one and precisely one link from companies to contacts flagging a primary
Say I have the following table:
TABLE: product
============================================================
| product_id | name | invoice_price | msrp |
------------------------------------------------------------
| 1 | Widget 1 | 10.00 | 15.00 |
------------------------------------------------------------
| 2 | Widget 2 | 8.00 | 12.00 |
------------------------------------------------------------
In this model, product_id is the PK and is referenced by a number of other tables.
I have a requirement that each row be unique. In the example about, a row is defined to be the name, invoice_price, and msrp columns. (Different tables may have varying definitions of which columns define a "row".)
QUESTIONS:
In the example above, should I make name, invoice_price, and msrp a composite key to guarantee uniqueness of each row?
If the answer to #1 is "yes", this would mean that the current PK, product_id, would not be defined as a key; rather, it would be just an auto-incrementing column. Would that be enough for other tables to use to create relationships to specific rows in the product table?
Note that in some cases, the table may have 10 or more columns that need to be unique. That'll be a lot of columns defining a composite key! Is that a bad thing?
I'm trying to decide if I should try to enforce such uniqueness in the database tier or the application tier. I feel I should do this in the database level, but I am concerned that there may be unintended side effects of using a non-key as a FK or having so many columns define a composite key.
When you have a lot of columns that you need to create a unique key across, create your own "key" using the data from the columns as the source. This would mean creating the key in the application layer, but the database would "enforce" the uniqueness. A simple method would be to use the md5 hash of all the sets of data for the record as your unique key. Then you just have a single piece of data you need to use in relations.
md5 is not guaranteed to be unique, but it may be good enough for your needs.
First off, your intuition to do it in the DB layer is correct if you can do it easily. This means even if your application logic changes, your DB constraints are still valid, lowering the chance of bugs.
But, are you sure you want uniqueness on that? I could easily see the same widget having different prices, say for sale items or what not.
I would recommend against enforcing uniqueness unless there's a real reason to.
You might have something like this (obvoiusly, don't use * in production code)
# get the lowest price for an item that's currently active
select *
from product p
where p.name = "widget 1" # a non-primary index on product.name would be advised
and p.active
order-by sale_price ascending
limit 1
You can define composite primary keys and also unique indexes. As long as your requirement is met, defining composite unique keys is not a bad design. Clearly, the more columns you add, the slower the process of updating the keys and searching the keys, but if the business requirement needs this, I don't think it is a negative as they have very optimized routines to do these.