Notes on Assignment Seven: Implementing I/O Routines

Even though I gave you folks a quiz to help ensure that you thought about some of the more complex issues, many of you failed to consider all of those issues. Many of you also ignored problems mentioned in the problem summary.

Utility

• Some of you still failed to provide a peek() or unread() method.
• Some of you failed to provide both left- and right-justification (both of which are useful).
• Some of you were clever enough to separate appending from writing.
• It was dismaying that so many of you simply did the equivalent of a readString and then a conversion to a separate type. A better implementation would have stopped when you saw the first inappropriate character.
• A user of these classes might assume that it would be reasonable to "invert" functions to read the result of a program. E.g., if I call writeXXX and then writeYYY, in generating output, shouldn't I be able to use readXXX and readYYY to read them in?
• Most of you limited the formats for input to decimal. What if I wanted hex or octal?

Ambiguity

• I was sorry to see that some of you still left it unclear as to where the cursor was after a failed read.

Coding Practices

• Many of you still need to work on your commenting style.
• Although most of you needed to skip whitespace in each of your routines, few bothered to write a separate utility method.

Re-Implementation

There was a wide variation in difficulty of implementation of the other group's classes. Some of you created very similar classes, which made implementation nearly trivial. Others of you found it literaly impossible to implement the other group's class using your own class. I tried to be sympathetic in the later case.

Testing

Many of you could have caught errors if you'd done better (any?) testing. I would have particularly liked to have seen testing of your implementations of the other group's classes.

I saw a variety of testing methods. Some of you took Bentley's advice to heart and built interactive test programs. Others of you tried to develop more comprehensive tests. Given Bentley's illustration of the use of loops in testing, I was sorry to see that few of you did such testing.

Misc.

A few groups put their pages directly on the web, which made access particularly nice and easy. I'd encourage the rest of you to do so for the remainder of your assignments (okay ... for your project).