Database queries pattern
Hello, new poster here
I'm often creating programs with database access. The way I do it now is to have a class for each type. For example, let's say I'm creating a job database. Then I could have a class called users, jobs, companies, and so on. Each of these will have some fields in them. Then I would implement create, read, update and delete (CRUD) functions for each class.
My question is the following. Since the code for each CRUD-implementation is roughly the same, just with the table and the field names changed, would it be a good idea to create a single implementation for the CRUD functions, and then extend this to each class.
One way I guess I could do this, is by including the database table name in each class, and a list of all field names in the class, then write generic CRUD fucntions, that loop through a list of field names.
The benefit would be less maintenance, faster to create new classes, generic handling of each class, faster to implement new functions (like fx. a find function), however I guess i loose some flexebility. But of course I could just implement CRUD functions on single classes if customization is needed by overriding the fudntions.
Is this a good or bad idea, or am I missing a crucial problem? Should it be implemented in another way than my idea with the list? How do you guys handle multiple classes with similar CRUD-functions?
I'm often creating programs with database access. The way I do it now is to have a class for each type. For example, let's say I'm creating a job database. Then I could have a class called users, jobs, companies, and so on. Each of these will have some fields in them. Then I would implement create, read, update and delete (CRUD) functions for each class.
My question is the following. Since the code for each CRUD-implementation is roughly the same, just with the table and the field names changed, would it be a good idea to create a single implementation for the CRUD functions, and then extend this to each class.
One way I guess I could do this, is by including the database table name in each class, and a list of all field names in the class, then write generic CRUD fucntions, that loop through a list of field names.
The benefit would be less maintenance, faster to create new classes, generic handling of each class, faster to implement new functions (like fx. a find function), however I guess i loose some flexebility. But of course I could just implement CRUD functions on single classes if customization is needed by overriding the fudntions.
Is this a good or bad idea, or am I missing a crucial problem? Should it be implemented in another way than my idea with the list? How do you guys handle multiple classes with similar CRUD-functions?
closed account (48T7M4Gy)
If it's a relational database then the normal SQL functionality and database connectivity or whatever the latest in ADO/ODBC is, is probably not worth tinkering with.
If you have an object oriented database the idea isn't a bad one provided you organise inheritance and the like in a rigorous and organised way.
However, to make that generic would be 'interesting' to say the least.
The data model is obviously key - <list>'s, <vector>s, <tuple>'s, <maps>'s ...
If you have an object oriented database the idea isn't a bad one provided you organise inheritance and the like in a rigorous and organised way.
However, to make that generic would be 'interesting' to say the least.
The data model is obviously key - <list>'s, <vector>s, <tuple>'s, <maps>'s ...
Yeah I see. I thought i could make a list of references to all variables that should be update in the database, but of course I can't do that because they are all of different types. I guess I can't make a list of pointers, where the pointers point to variables of different types? Maybe I could create a class that could hold, a int, double, string, datetime and so on, and make a list of that class. Then when i call that class to get the pointer it would return the one that is non-empty.
So there is no obvious way to do what I want? All people working with this kind of object oriented database write the CRUD-functions for every single class even though it is more or less the same each time?
So there is no obvious way to do what I want? All people working with this kind of object oriented database write the CRUD-functions for every single class even though it is more or less the same each time?
closed account (48T7M4Gy)
| Maybe I could create a class that could hold, a int, double, string, datetime and so on, and make a list of that class |
| All people working with this kind of object oriented database write the CRUD-functions for every single class even though it is more or less the same each time? |
There is plenty of reference material on OO databases.
As an observation though, you seem to be conflating the data types (classes) with the data model (<vector>'s etc). Both are easy to implement with an OO management program you design to do CRUD.
Keep in mind the create-C part is simple a constructor and add to a <vector> say. It's no 'bigly'. The others in CRUD aren't insurmountable ether.
| So there is no obvious way to do what I want? |
https://www.webtoolkit.eu/wt/doc/tutorial/dbo.html
Since 'Action' can be anything, you may even extend the behavior to compare two datasets. You may use your own 'field' function.
Actually I think the approach where a certain templated function processes all fields is good.
Topic archived. No new replies allowed.