Today, in this final post of the semester for CS-343 I will be examining the JPA architecture. This is something I have been curious about after browsing articles on DZone for the past couple of weeks and seeing this topic pop up a lot, luckily they made a good introduction post this week. From the article: “The Java Persistence API is the Java standard for mapping Java objects to a relational database”, this is something that I have been wondering how to do since we started working with the REST APIs and our projects that have been using Java HashMaps for makeshift databases instead of something more permanent like SQL. This article briefly introduces and explains the functionality of the six key parts of the architecture for JPA. The article gives a diagram for these six units and the multiplicity of their relationships. The article then gives a simple example of mapping a simple Java POJO Student class into data for a database, using the student ID as the primary key. I was surprised at how easy this translation was, all it includes is a couple of tags for columns above the instance variables that designate the database column, and which one is the primary key to generate in the database. The article then creates an application class that uses the various entity classes to store a newly created student into a database. I liked the simplicity of the article as most introductory articles are on this website for what are advanced programming topics, but my biggest criticism is that this article has brief definitions for many of the elements of JPA and then links to an external site that further elaborates on them. I wished that the author put a bit more time and consolidated all of this information into one article for this site, instead of having a bunch of separate webpages on his own website. I also wish he showed in this post getting data from the database and translating it back into Java and performing manipulations on this data with methods. Overall it was a good introduction to a very useful topic and one that I would like to further look at. Going forward, I am much in favor of using this way of storing data as opposed to simple Java in-memory databases for entire applications like we did in our projects.
Today I am looking at another design pattern. This one is the command design pattern, which is yet another Gang of Four design. They classify it as a behavioral pattern with object as the scope (GoF 223). I wanted to explore this one as reading through the different pattern descriptions in the Gang of Four book, this peaked my interested by its ability to separate requests from objects. The article gives a good summary of the idea of the pattern with a real-life example of a restaurant with a server taking an order from a customer and handing it off to a chef. It then further breaks down the pattern with the usual UML diagram and a helpful sequence diagram that shows the order in which the classes perform the pattern. I found this sequence diagram, along with the comments in the example program with code that show which class matches with which part of the diagrams really useful in this example, as the pattern goes through a couple of classes just to call a basic function on a simple object. Although this pattern does seem complex at first, it has a nice simplicity once you understand what all the classes are doing, and once you get the base created adding more functions is as simple as adding more command classes. The article then creates a simple example with Java of the command pattern using the various classes to switch a light bulb object on and off. I do like the idea of the pattern and its particular implementation in this example. It nicely breaks down requests into their own separate command objects that gives much greater control over requests across a program. I agree that the ability this pattern gives to create a log of function calls and add an ability to undo all functions called on an object is very helpful. I also agree with the author that it can get messy if you have a lot of functions that need to be implemented. As this pattern calls for a separate command class to be made for each function or method performed on an object, this can quickly add up depending on how many methods you need in your program. In the future, I will definitely remember this pattern and its useful ability to separate commands from the objects it performs them on.
Today I will be looking at an article that examines five basic guidelines when creating a REST API program. I thought that after spending over a month on our final project creating a REST API backend it would be good to review some good practices for creating these REST APIs. The article gives five basic guidelines to follow and a brief background on each of the guidelines. The five guidelines are: name and case conventions for URIs, the different HTTP methods, HTTP headers, query parameters, and status codes. Some of these topics were a refresher from what we’ve already been taught, such as some of the HTTP status codes, and others were new such as the query parameters or case conventions. I especially thought the name and case conventions for the URIs was interesting. I don’t think I have reviewed naming conventions in coding in a couple of years, I usually use camel case for everything since I mostly write in Java. But I have been wondering what was the best practice for endpoints since I remembered code with HTML being more case sensitive. Also, with using different naming schemes instead of just using verbs that describe the method as in Java. I will keep in mind going forward to use the spinal-case method when creating endpoints. The HTTP methods section was fairly straightforward with what I already know, and the use cases for most of them are the same, but it does have a couple of additional methods I haven’t seen before such as HEAD or OPTIONS. The HTTP headers section was interesting, giving definitions and names for the different types of headers you use for different requests. The query parameters section was the most interesting and something that I haven’t come across yet working with REST. The status codes were familiar too, I have mostly used OK, CREATED, and NOT FOUND in my endpoints. Although I will start using NO CONTENT when performing a deletion endpoint instead of just 200-OK, as this seems to be the more proper response code. The same with using a 400 error of BAD REQUEST instead of just giving everything a 404-NOT FOUND error when an invalid request is made or an item is not available. In the future when working with REST I will keep in mind these simple guidelines from this article, especially in regards to case conventions for endpoints and with using appropriate status messages.
Today I learned more about the Iterator Design Pattern and how it specifically applies to Java. The reason for choosing this topic is that during my project and talking with others the Iterator has come up a number of times throughout this semester. Now, I already knew about the Iterator in Java from taking Data Structures, but I never realized it was a design pattern before this course. The Iterator is actually a Gang of Four design pattern and it is classified as a behavioral pattern with object as the scope (GoF 257). The article gives a short summary of the iterator pattern and the reason it is used in Java. It is basically a way of accessing elements in a collection in order without having to know about the internal representation of the structure, a nice way of using abstraction. The author then gives a nice simple example in Java of how to use the Iterator pattern to access a collection of elements. He does this by creating a basic shape POJO with an id int and String name and stores them in another class that is an array of shapes. He then creates an Iterator class that defines the next element and if there is another element after the current one. This may be my favorite design pattern I’ve seen yet. It is simple to implement, but yet effective at its purpose, especially the way this article showed its implementation. I really like that once you’ve created an iterator for a type, all you need to do is pass in a collection of elements to it in order to process it. When I was working on the backend for my project I realized the need for something like an Iterator, especially with all the endpoints that were looping through the whole database. I wish I had implemented this so that I did not have to keep rewriting conditions to check that the data wasn’t out of bounds. In the future when I create my own programs with collections of elements I will make sure to implement an Iterator so that I can easily cycle through the data without having to worry about how to do it or constantly bounds checking the collection.
As we are nearing the end of the semester and about to finish our projects, I’ve been thinking more about the documentation process for the different parts including the frontend and backend. With this in mind, today I found a great topic about documentation generation for REST APIs with a tool called Swagger. I’ve never heard of Swagger before today, but it is a useful framework that allows for testing, documentation, and other useful features for building APIs. This post shows a quick little tutorial of how to implement Swagger for documentation generation with a Spring Boot project that is very similar to my project and the example REST API order system we have been using. It is a relatively simple process that includes adding the necessary dependencies for Swagger and adding a controller class for it, then all you have to do is just add the necessary documentation statements for each controller and requests within the controllers. The end result is a very nice-looking HTML page that displays a well formatted layout which includes the documentation for your API backend with a graphical display for each request and all the information associated with the request such as body and return information. Now as I was reading this article and this new way of creating documentation, I was comparing it to the way we’ve been doing it so far with a simple table written in Markdown for all of our API endpoints. The output of doing it in Markdown was nice but writing it was a tedious task with the formatting of the table. I much prefer the simplicity that Swagger allows you when adding a new endpoint. I also like the final product that Swagger produces a lot more than the simple Markdown document. In the future if I am creating another Spring Boot project, I am going to try to use Swagger from the start for documentation instead of using a Markdown document with a table for a basic readme as it appears that using Swagger makes adding new endpoints much less tedious with formatting. I would also like to try to add this to my current project if there is enough time, and also see if it is possible to use it with Angular too.
The article that I read this week is about processing JSON data with Jackson. I thought this would be a good article to read after using JSON data a lot this week for processing input and output in our final projects, and since Spring Boot uses Jackson as the way for processing JSON. I wanted to learn more about how this works, especially after getting some errors when trying to pass in certain data types this week (particularly Calendar objects). This post goes over how to read and write JSON data to and from Java objects using Jackson for data binding. It does this by creating an example POJO to use for input and output. It creates a basic employee with the fields of name, ID number, address and other typical fields, it also has an object within an object by using an address within Employee that contains a street, city, and zip code. The article creates an example of this in a JSON input file and creates the necessary Java classes and then implements the Jackson methods (such as ObjectMapper) for databinding and outputting Java input as a JSON file.
I think that it was interesting to see how to do this with the Jackson implementation as a seemingly more proper method of converting JSON to Java objects and the other way around. Especially after spending the past week creating and getting our project to pass similar data back and forth between JSON and Java. It does seem out of the ordinary to me that both of the example classes don’t use constructors, instead using just set methods to create the object, but that’s how this implementation is supposed to work with Jackson. I particularly liked the Tree Model implementation in the article and was not aware that this was a way of processing JSON data. This article has definitely made me think more about the different ways of processing JSON data with web applications and REST APIs and the best practices to use when doing this, especially with larger, serious implementations for applications. If time permits for our final project, I would like to try and do a similar implementation in this article for our JSON processing. By doing this, it would make adding new objects to the database a lot cleaner (especially without needing a constructor) in the implementation than it currently is.
This post on DZone talks about the abstract factory design pattern and gives an example implementation in Java using geometric shapes. This pattern is similar to the simple factory with the idea of constructing objects in factories instead of just doing so in a client class. It differs in that this abstract version allows you to have an abstract factory base that allows multiple implementations for more specific versions of the same original type of object. It also differs in that you actually create an instance of a factory object instead of just creating different objects within the factory class as in the simple factory.
I like the concept of this pattern more than just having a simple class that creates multiple instances of different objects such as the simple factory. I also like how the design allows you to have multiple types of objects that can split off into different more specific types, such as how the example Java implementation has 2D shapes and 3D shape types and factories for each kind. The design appears to be efficient, especially in the implementation example, only creating a factory for a type of object when it matches a specific type in the client call. Like the other factory pattern, you can also easily implement other design patterns for the object itself such as a strategy or singleton, which further would improve the final outcome. Another aspect of this pattern that I like is that the client itself is not creating the objects, it just calls the factory get method from a provider class that sits between the factory and the client.
I definitely like this pattern and will certainly consider using it the next time I have to create a program with many different variations of the same objects such as shapes or ducks as seen in previous programming examples. It will be especially useful to use this design if I am trying to type check the objects from user input to make sure they are trying to create a valid type of object with the factory. Overall, I am finding that as I read more articles about design patterns, especially for many objects of the same base, I am gaining a better understanding of how to maximize the program efficiency with one or multiple design patterns.