tP: Constraints

The product should be targeting users who can type fast and prefer typing over other means of input.
Reason: to increase comparability of products, and to make feature evaluation easier for peer evaluators.

The product should be for a single user i.e. (not a multi-user product).
Not allowed: Application running in a shared computer and different people using it at different times.
Not allowed: The data file created by one user being accessed by another user during regular operations (e.g., through a shared file storage mechanism).
Reason: multi-user systems are hard to test, which is unfair for peer testers who will be graded based on the number of bugs they find.

The product needs to be developed in a breadth-first incremental manner over the project duration. While it is fine to do less in some weeks and more in other weeks, a reasonably consistent delivery rate is expected. For example, it is not acceptable to do the entire project over the recess week and do almost nothing for the remainder of the semester.
Reasons: 1. To simulate a real project where you have to work on a code base over a long period, possibly with breaks in the middle. 2. To learn how to deliver big features in small increments.

The data should be stored locally and should be in a human editable text file.
Reason: To allow advanced users to manipulate the data by editing the data file.

Do not use a Database Management System e.g., MySQLDBMS to store data.
Reason: Using a DBMS to store data will reduce the room to apply OOP techniques to manage data. It is true that most real world systems use a DBMS, but given the small size of this project, we need to optimize it for CS2103/T module learning outcomes; covering DBMS-related topics will have to be left to database modules or level 3 project modules.

The software should follow the Object-oriented paradigm primarily (but you are allowed to mix in a bit other styles when justifiable).
Reason: For you to practice using OOP in a non-trivial project.

The software should work on the Windows, Linux, and OS-X platforms. Even if you are unable to manually test the app on all three platforms, deliberately avoid using OS-dependent libraries and OS-specific features.
Reason: Peer testers should be able to use any of these platforms.

The software should work on a computer that has version 11 of Java i.e., no other Java version installed.

The software should work without requiring an installer.
Reason: Testers may not want to install your product on their computer.

The software should not depend on your own remote server.
Reason: Anyone should be able to use/test your app any time, even after the semester is over.

The use of third-party frameworks/libraries/services is allowed but only if they,

• are free, open-source (this doesn't apply to services), and have permissive license terms (E.g., trial version of libraries that require purchase after N days/uses are not allowed).
• do not require any installation by the user of your software.
  In case of services, requiring the user to create an account on a third-party service is strongly discouraged as it can result in your product for 'low testability'.
• do not violate other constraints.

and is subjected to prior approval by the teaching team.
Reason: We will not allow third-party software that can interfere with the learning objectives of the module.

Please post in the forum your request to use a third-party libraries/services before you start using the library. Once a specific software has been approved for one team, other teams may use it without requesting permission again.
Reason: The whole class should know which external software are used by others so that they can do the same if they wish to.

Package everything into a single JAR file.
Reason: Requiring the users to download more files can cause complications in the final peer testing session.

If you are unable to package everything into a JAR file, package the JAR file and other files needed into a single zip file.

The file sizes of the deliverables should be reasonable and not exceed the limits given below.

• Product (i.e., the JAR/ZIP file): 100MB (Some third-party software -- e.g., Stanford NLP library, certain graphics libraries -- can cause you to exceed this limit)

• Documents (i.e., PDF files): 15MB/file (Not following the recommended method of converting to PDF format can cause big PDF files. Another cause is using unnecessarily high resolution images for screenshots).

In addition, do ensure that the final JAR/PDF files are not bloated unnecessarily. Such bloat can be reported as a bug. Some suggestions:

• Check if the the assets (e.g., images, audio, data) included in the JAR files are all strictly necessary and the quality is not unnecessarily high (e.g., images with higher resolution than necessary).
• Check if the third-party libraries in the JAR file are strictly necessary or whether they have lighter versions that are still enough for your purpose.
• Using JavaFX WebView allows you to display a Web page within your application but it adds about 70MB to your JAR file. If you decide to use that library, ensure the benefit is worth the increase in size.

The DG and UG should be PDF-friendly. Don't use expandable panels, embedded videos, animated GIFs etc.

Reason: The UG and DG used in the final grading will be in PDF format:

It is OK to use a reliable public API e.g., Google search but we recommend that you have a fallback mechanism (e.g., able to load data using a data file if the network is down).
Reason: During the mass peer-testing session, the network access can be intermittent due to high load. If your feature cannot be tested due to lack of Internet, that will have to be counted as a major bug, to be fair to those whose app is being tested and bugs found being penalized.
If you use NUS data (e.g., scrape data from an NUS website), please work with NUS IT directly to get their approval first. Even well-intentioned use of NUS data without approval can get you into serious trouble (has happened before). The teaching team will not be able to get approval for you as the use of NUS data is not a module requirement.

Avoid implementing hard-to-test (both for manual testing as well as automated testing) features or features that make your product hard-to-test.
Reason: testability is a grading criterion. If you choose to implement such a feature, you will need to spend an extra effort to reach an acceptable level of testability. Here are some examples of features that are hard-to-test:

• Features that depend heavily on remote APIs: Those APIs can block your access if they mistake your automated tests as a bot attack. Some remote APIs require setting up accounts, keys, login etc, that will irritate the testers of your product and give a low rating to the testability of your work.
• Audio-related features: The peer testing of your product is done under exam conditions where it is not appropriate to play audio.
• Features that require creating user accounts, login, logout etc.

Following from the Constraint-Typing-Preferred, if the app is optimized for the target user (graded under the product design criterion), a user who can type fast should be able to accomplish most tasks faster via CLI, compared to a hypothetical GUI-only version of the app. For example, adding a new entity via the CLI should be faster than entering the same data through a GUI form.
Therefore, the input to the app needs to be primarily CLI. The GUI is used primarily to give visual feedback to the user. This does not mean the app need to have a text UI (CLI is not the same as text UI) or it needs to work in a command console -- it simply means the input to the app should be a text command, which can even be entered using a GUI (similar to how AB3 does it). While we don't prohibit non-CLI inputs (e.g., clicking of a button), note that such inputs will reduce the suitability of the product to target users. Therefore, give CLI alternatives to mouse/GUI inputs, if applicable.
Also keep in mind:

• Regular typing is usually faster than using key combinations.
• typing the full command and hitting ENTER will complete the taskOne-shot commands are faster over prompting the user to input one parameter at a timemulti-step commands. If you provide a multi-step command to help new users, it is recommended that you also provide a one-shot equivalent for regular/expert users.