|By Ben Forta||
|October 18, 1999 12:00 AM EDT||
No one wants to write buggy code, at least no one I choose to know. Bugs are annoying, bugs are embarrassing. And bugs can cost you (and your clients) lots of time and money.
Bugfree code is the ideal all developers strive for - at least should strive for - but it's a lofty goal not easily attained.
To write bugfree code it's important to understand how bugs are introduced. I'd like to explore what I believe to be one of the primary causes for the introduction of bugs into your code: failure to expect the unexpected. While many of these ideas apply to application development in general, the positioning of this column relates specifically to ColdFusion.
Flow? What Flow?
Applications are designed with a particular program flow in mind. Users start at point A, go to point B, then to point C. Program flow is an important part of any application, and when developers write they anticipate a particular program flow.
Here are some examples.
In each of these examples there's a logical starting point, then a series of steps. The search dialog must be displayed before the search can be performed, users must select items from the product catalog before the order can be processed, and login information must be provided before access can be authenticated.
In traditional application development, programmers had total control over program flow. Users had no way to access screens out of order, nor could they start from a screen other than the one they were supposed to start at. But Web applications, for better or worse, behave differently. As every Web page has a unique address (its URL), it's indeed possible for users to execute pages out of order, just as it's possible for them to start from the wrong page.
How could this occur? There are several ways. Users could bookmark pages directly, search engines might index pages below your root, and newer browsers feature an auto-fill option that completes URLs for users but frequently uses the most recently visited page in the site (often the wrong page).
To understand this better, look at the following code:
<!--- Perform search --->
<CFQUERY DATASOURCE="ds" NAME="search">
SELECT title FROM books
WHERE title LIKE %#FORM.title#%'
<!--- Create page --->
<!--- Display search results --->
Actually, this is fairly typical ColdFusion code (it's even commented). It performs a simple database search and displays the results. The search itself is driven by a form that contains a field named "title"
And that's what's wrong. The code makes two dangerous assumptions - that the page will be called from a form and that the form contains a field named "title". If either assumption turns out to be incorrect, ColdFusion throws an error because #FORM.title# would refer to a variable that didn't exist.
What's the solution? There are a couple of things you can and should do. The first is always check for the existence of variables before using them, initializing them with default values if needed. The ColdFusion <CFPARAM> tag is very useful for this, and good programming practices demand that every variable be initialized this way so they always exist (either as submitted FORM or URL values, or as locally created variables). If the following code were inserted above the code example, the page would always execute correctly, regardless of how it was invoked and what fields were passed:
<CFPARAM NAME="FORM.title" DEFAULT="">
Sometimes you may not want to use default values. For example, in the preceding code you may not want to display the entire contents of the table if the page is executed directly. If a user ends up on this page without having filled in the form to perform the search, you'd want to send them to the search page instead. Here's one way you could do this (assuming the search page was named search.cfm):
<!--- Is form field present? --->
<CFIF NOT IsDefined("FORM.title")>
<!--- Not present, redirect to search screen --->
With this snippet at the top of the page your code is now safe. If users try to execute the page directly, you'll programmatically redirect them to where they really should be.
The truth is, every page on your site should be written so that it's safe to execute directly. That way you'll never make assumptions about program flow, and your code won't break when the unexpected occurs.
Are you up for a challenge? How about writing code that uses <CFDIRECTORY> to traverse your code tree to find all CFM pages, then looping through each one, executing it via <CFHTTP> to see which ones throw errors and which work. You could call this code daily (or weekly or monthly) and e-mail yourself the results so you'll know as soon as possible if things might break.
Changes in Site Usage
Another phenomenon that's part of Web site use is the frequency with which that use changes. Programmers anticipate that their site will be used in one way, but, inevitably, users find another way to use it. That's not a bad thing in and of itself, but it does present an interesting problem.
All applications, not just Web-based applications, are designed so that particular parts perform better than others. This is usually because certain parts of the application are more critical and get more use, so the time and effort needed to improve and enhance the site tends to be best spent on these parts.
The application hums along nicely, handling the load thrown at it, and everything works well until those pesky users start using your site in ways you didn't expect. Suddenly the efficient and highly optimized parts are being executed less and less, and the less fine-tuned (I'm being generous) parts start to buckle under the growing load.
This is actually a very common problem. Some of the largest and most impressive sites on the Internet have fallen victim to it.
The good news is that ColdFusion provides you with an invaluable (and frequently overlooked) tool you can use to identify these potential trouble spots. The ColdFusion Administrator (starting in version 4) features a checkbox that, when checked, allows you to log page requests that take longer than a specified amount of time.
Again, the key here is to expect the unexpected and be ready when it occurs. Individual log entries aren't necessarily indicative of a problem, but repeated entries most definitely are. I'd strongly advise all site administrators to enable this option, even on production sites. Determine what the "normal" response time should be (and then pad that number a little) and have ColdFusion log all page requests that take longer. If specific pages start appearing in the log repeatedly, you'll know where to spend your fine-tuning and performance-enhancing efforts.
Up for another challenge? Write a scheduled event that checks this log file daily for new entries. If it finds any, it should e-mail them to you so you'll know about them immediately.
When Bad Things Happen
You've cleaned up your site. You no longer expect specific program flow, nor do you expect specific usage patterns. That's great, but it's not enough.
The other big trouble spot is reliance on other systems and technologies. Whether it's database integration, execution of third-party objects and components, or interaction with other Internet protocols (like HTTP, FTP and NNTP), the more you rely on other systems and technologies, the more room there is for something to go wrong.
Whether it's database failure, the inability to connect to a specific host, or errors and problems in calls resources, they're all your problem because to the end user your site is broken. It's not an ideal world out there; bad things do happen and you'll be blamed. And you're not going to escape this one anytime soon.
So what do you do? Again, ColdFusion (version 4 or later) provides a tool to respond to these situations - try catch error handling. While full coverage of try catch is beyond the scope of this column (I'll devote an entire column to error handling in the future), here's the basic idea.
Try catch error handling allows you to trap errors in your code, then pick up the processing elsewhere when an error occurs. The basic page layout looks like this:
... page goes here ...
... error processing goes here ...
By wrapping an entire page between <CFTRY> and </CFTRY> tags, any error that occurs within that page will be trapped. As soon as an error is caught, processing will be taken over by the catch block (the code between <CFCATCH> and </CFCATCH>). There's no limit to what you can place in the catch block-you can change variables, redirect requests, generate e-mail, write logs and even try to fix error conditions.
Experienced developers take this one step further by nesting try catch error handling. Generic error handling is implemented at the page level, and additional error handling is implemented around specific features or functions. This type of implementation provides the greatest level of control and allows developers to not just expect the unexpected, but also to handle the unexpected.
The Internet is a highly dynamic and configurable environment. With all that flexibility and power comes the need for greater caution and planning to ensure that applications don't suddenly break. Bugfree code is a wonderful ideal, but for most of us, writing good code that doesn't break in unexpected scenarios is as close as we'll get to attaining it.
The bottom line is that as developers writing code for this new environment, we must all expect the unexpected- because it's going to happen and usually at the least opportune time.
- Where Are RIA Technologies Headed in 2008?
- The Next Programming Models, RIAs and Composite Applications
- AJAX World RIA Conference & Expo Kicks Off in New York City
- Constructing an Application with Flash Forms from the Ground Up
- Building a Zip Code Proximity Search with ColdFusion
- Personal Branding Checklist
- CFEclipse: The Developer's IDE, Eclipse For ColdFusion
- Has the Technology Bounceback Begun?
- Cloud People: A Who's Who of Cloud Computing
- i-Technology Viewpoint: We Need Not More Frameworks, But Better Programmers
- Adobe Flex 2: Advanced DataGrid
- Web Services Using ColdFusion and Apache CXF