DeepCode’s Top Suggestions #1: Java Date (This one made me dizzy)

Hey,

you might know, DeepCode uses thousands of open source repos to train our engine. We asked the engine team to provide some stats on the findings. On the top suggestions from our engine, we want to introduce and give some background in this series of blog articles. And this one made me dizzy as we are looking back on like 70 years of software engineering and still have problems with — drumroll — date formats…

Language: Java
Defect: Date Data Format (Category General 1)
Diagnose: Ambiguous date-time output formats (e.g. 12h output without am/pm suffix)

Background:

Quick, what is the difference between the date format strings MM-DD-YYYY and mm-dd-yyyy or MM-dd-yyyy ?? I can tell you, they will all have the same result when you call them at the right moment. But obviously, they are fundamentally different. Well, let us shed some light…

Java had a bad start regarding its date classes. java.util.Date had serious design flaws when it was introduced which led to lots of confusion. A Date instance in Java is actually not a date but a moment in time, therefore (1) it has no time zone, (2) no format, (3) no calendar system. I suggest this valuable blog post for the full story. After years, the Java community acted and introduced new classes. Still, there are lots of traps (for example, java.text.SimpleDateFormat is not thread-safe while java.time.format.DateTimeFormatter is) but for now, let us focus the most common mistake which is the one flagged by DeepCode:

• Using mm for months and/or MM for minutes (wrong!).
• Using hh for “hour of the day” when really HH was intended. HH ranges from 0 to 23 while h ranges from 1 to 12 and is mostly used as single character. It needs the AM/PM information or it is ambiguous.
• Using YYYY for year. It is meant to be used in conjunction with “week of the year” and can lead to unexpected results in the first and last week of a year. Normally, you want to use yyyy .
• Using DD for “day of the month” but in reality, it means “day of the year”

Make sure to use the correct pattern characters. As a reference in Java, the following applies.

Note: This is Java. Do not simply expect this to be the same elsewhere. Always check the documentation.

This is not exhaustive on problems around dates and times. We could talk about the difference between UTC offset and timezones, the problems around timezone abbreviations (is BST British Summer Time or British Standard Time or rather Bougainville Standard Time (No, I did not make this up)? Well, who knows), or different calendars in different locales.

To provide the answer to our little puzzle above, you probably almost always want MM-dd-yyyy .

CU

0xff