quartz/content/notes/18-advanced-sql-1.md

title	aliases	tags	sr-due	sr-interval	sr-ease
18-advanced-SQL		info201 lecture	2022-05-15	3	250

varchar usually bigger than you think

CRUD

• insert adds a row
• select retrieves rows from the table
  • ouput can be "saved" as a view
  • changes to the underlying table also chagnes the view
• update modifies rows
• delete removes rows
  • test as a select statement first

SQL DAO programming

We want to miniminse load on sql as connecting to database is expensive.

Optimisations:

• prefer muultiple row operations
• connection pools (keep connections "alive")
• reuse prepared statements (reduce unnecessary SQL parsing)
• consider combining queries to reduce round trips
• batched requests

Follow Optimistic approach

• assume operatios will succeed (no pre checking)
• handle errors with exception handling
• consider using merge if available

Merge

aka replace etc.

• update if they exist other wise insert

JDBC

Java framework for interacting with sql databases.

plaform and DBMS independent

• driver provided by DBMS vendors
• same Java code will work with any DBMC

key concepts

• connections and connection pools
• sql strings and prepared statements
• result sets
• transactions
• batched requrests

JDBI

• bette version of JDBI
  • layered on top of JDBC
  • better APIs
  • less code
• simple class <-> table mapping
• flexible plugin architecture

Transactions

interaction between two entities

• follow explicit or implied forms
• usually involves exchange on resources
• may require several steps
• often considered a single unit

In data bases

• group of db operations is considered a single logical unit
  • transfer (read and update)
  • recieve shipement (update accounts)
  • customer sale

transactions are all or nothing. (commit vs rollback)

ACID

• Atomic
  • all or nothing
  • operations should be related
• Consistent
  • transactions move dbs from one consistent state to another
    • "consistent" ⇒ all integrity rules are satisfied
  • db may be inconsistent during a transaction
    • require defferable constraints
• Isolated
  • concurrent transactios shouldn't interfere with each other
  • ideally behave as if other transactions dont exist
  • read committed isolation
    • uncommited changes are visible to other transactions
  • require some form of concrrency mangement (e.g., locking)
  • 
• Durability
  • once a transaction is commited it is permanent
  • uncommited transaction dont survive crashes
  • auto rollback of uncommitted transaction

commit and rollback

• changes are made to "live" data
• commit makes database changes permanent
• rollback removes all changes since that transaction start

Transaction in Java

default to auto commit.

• each statement is a separate transaction
• if transaction has multiple statements
  • disable auto commit
  • you must manage commit and rollback yourself

Select

• select <- wahat
• from <- from where
• where <- filter
• group <- aggregation
• order <- order

distinct removes duplicate rows