I simply cannot fully grasp this concept no matter how hard I try. I'd love a few explanations!
For further actions, you may consider blocking this person and/or reporting abuse
I simply cannot fully grasp this concept no matter how hard I try. I'd love a few explanations!
For further actions, you may consider blocking this person and/or reporting abuse
JS -
Anuj Tyagi -
Abdur Rakib Rony -
Tingwei -
Top comments (10)
There are two ways to make a pancakes: with eggs, flour, baking powder and milk or with a pancake mix.
Think of a database view as pancake mix. The ingredients were already prepared before. You just have to mix and heat it.
If I'm not mistaken then, they sound like assigning a method name to an SQL function that was built previously.
Am I close?
Yes, you're getting close. I'll try to write some pseudo code so you can understand better:
Not assigning a function, but a SQL statement. In most cases a query.
Why not have this logic stored in the program itself, even as a simple string variable? What am I gaining from using this functionality?
Not challenging, 100% curious.
One nice thing about a function stored in SQL is that SQL can create an "execution plan" and store it as well, so it doesn't have to reparse and plan everything each time you submit the query. It speeds things up, especially with queries you use frequently.
For me, views often are kind of abstractions or interfaces. Especially when the view contains joins. It hides the details of the data model. Database programmers can even change those details while maintaining the SELECT clause of the view so for the application programmer the interface has not changed.
Hey Ben,
SQL views aren't just stored queries, it's a virtual table. You can make joins and select specific columns, granting permissions to different database users to it, instead of granting permission to every table. Views are also indexed, when talking about materialized views.
So by using SQL views you've got: security, performance and simplicity. Also, not every system is backed by a modern application layer.
These answers are partly true. Everything said so far is pretty correct, especially considering the open source community used MySQL for so long which only has this type of view. However, in postgres, you have the option of creating a materialized view which is everything said above but then the database runs the query and stores the data to disk so that running queries against it doesn't spend the time running again. 9 times out of 10 in postgres, you will want to use materialized views.
The benefits of basic views are pretty much barely anything at all. You can get the benefits of a pre-planned and stored query by using prepared statements (which, you probably already do automatically if you use a modern framework as this is a standard defense against SQL injection).
At the end of the day, all features, views included are tradeoffs. I rarely find basic views to be worth it (though others on this page clearly have found reason for it). I DO find that materialized views give me enough value for the added complexity to justify their use when needed. Note that materialized views must be refreshed as they are a kind of caching of your data. But, then again, I really find that views make the most sense on the reporting side of things where caching is a respected and legitimate pattern.
One thing I used to use views for was on an unknown schema to join everything I could find together into one enormous pile of data, so I could get an overview of what was going on quicker. I wouldn't use these monsters for anything apart from that though!
As far as I understand it in PostgreSQL, views are a snapshot of a table at a certain time defined by a query. Once you define a view you can call it as, essentially a read-only table.