- Ben Pan
- Frank Wong
Period: 3
We aim to implement a basic database management system (similar to Sqlite, Name: TBD) and a corresponding language (similar to SQL, Name: TBD).
The database will be stored locally. Multiple users will be able to access the database ("SELECT"), but only one will be able to make modifications (database will lock for a very short period of time) using our language.
| Technique | Description |
|---|---|
| Memory Allocation | Store query return values, vectors |
| Working with files | Access and modify database files and tables |
| Processes (fork) | Testing concurrency for multiple users |
| Sockets | Handle multiple connections from users |
| Semaphores | Prevent multiple users from modifying the same data concurrently |
| Finding information about files | Get information about database files and tables |
-
Phase 1a: Ben & Frank. Straightforward implementation, so will split work evenly
-
Phase 1b: Ben
-
Phase 2: Frank
-
Phase 3 and 4: Ben & Frank since these are advanced features that require further research together
We will start with basic data structures like arrays and linked lists for implementing the database tables/files.
Parsing might require the use of a tree.
If time allows, we will implement a B-Tree which allows search, deletions, and insertions to be done in O(log n) time. For now, we will use linear search and arrays/linked lists instead.
Another feature we want to implement is the relational model.
- Local single process database editor, no sockets or semaphores yet
- Allow basic fixed-form queries (eg.
INSERT INTO table_name VALUES (a, b, c), no language parsing, multiple arguments etc.) yet - Dynamic 1D and 2D array structs (likely void * vectors)
- Read / write from file with custom format for persistence
- Table likely restricted in terms of data types (integers only)
- Semaphore for file access management (no writing by multiple processes at once)
- Better database schemas to allow for multiple data types (eg. strings, floating-point numbers)
- Export to CSV file functionality to allow us to see the database with a standard format
- Forking server & socket-based system to allow for multiple clients to connect to database server
- Different client code that allows you to connect to a server
- Better parsing code to allow for more complicated queries (eg. AND in conditions, wildcard support beyond
SELECT * FROM table_name) - Client authentication, eg. require credentials for client to modify table
- Upgrade schema to support primary key / unique specifiers, basic relational model support
- Use data structures (eg. aforementioned B-Tree) to speed operations up
- Improve parsing algorithm if needed to support more edge cases / be quicker