Opinion
Computing Applications Kode Vicious

Forced Exception Handling

You can never discount the human element in programming.
Posted
  1. Dear KV,
  2. Dear Helpfully,
  3. Author
Forced Exception Handling, illustration

back to top  Dear KV,

I subscribe to “The Morning Paper,” a daily summary prepared by one person, Adrian Colyer, who curates research papers and sends them out to interested readers (https://blog.acolyer.org). Last fall he reviewed “Simple Testing Can Prevent Most Critical Failures: An Analysis of Production Failures in Distributed DataIntensive Systems” (https://blog.acolyer.org/2016/10/06/simple-testing-can-prevent-most-critical-failures/). It had some surprising results, including:

  • Almost all catastrophic failures (48 in total, or 92%) are the result of incorrect handling of nonfatal errors explicitly signaled in software;
  • Error handlers with TODO or FIXME in the comments. This example took down a 4,000-node production cluster; and
  • Error handlers that catch an abstract exception type (for example, Exception or Throwable in Java) and then take drastic action such as aborting the system. This example brought down a whole Hadoop Distributed File System (HDFS) cluster.

And the list went on from there.

I have been reading your Kode Vicious columns for quite a while, and as I read the review and then the paper itself, it looked like something you would be interested in, so I have sent along the link.

Helpfully Not in Error

Back to Top

Dear Helpfully,

Yes, KV also reads “The Morning Paper,” although he has to admit that he does not read everything that arrives in his inbox from that list. Of course, the paper you mention piqued my interest, and one of the things you did not point out is that it is actually a study of distributed systems failures. Now, how can we make programming harder? I know! Let’s take a problem on a single system and distribute it. Someday I would like to see a paper that tells us if problems in distributed systems increase along with the number of nodes, or the number of interconnections. Being an optimist, I can only imagine that it is N(N + 1) / 2, or worse.

I don’t think you pointed out this paper to KV just to hear me bang my head on my desk while thinking distributed systems, so let’s assume you’re asking the “Why?” question: “Why is it the case that 92% of the catastrophic failures in this paper are caused by a failure to handle nonfatal errors?”

Well, let’s see what else the paper had to say and then think about how software is actually implemented in the real world, rather than how we believe it ought to be implemented in the illusory world that management and marketing inhabit.

To get to the heart of why nonfatal errors might have led to fatal errors, we need look no further than this snippet from the paper: “This difference is likely because the Java compiler forces developers to catch all the checked exceptions; and a variety of errors are expected to occur in large distributed systems, and the developers program more defensively. However, we found they were often simply sloppy in handling these errors” (https://www.usenix.org/system/files/conference/osdi14/osdi14-paper-yuan.pdf).

Hopefully anyone who has been a professional programmer for more than a few days knows that many developers will always write the code they are most interested in, or pressured to deliver first, which is not the error-and exception-handling code, nor is it test code, nor documentation, the latter two of which I have already harangued readers about, ad nauseam. What management and the rest of the team want, is “the code,” and what most people see as “the code” is only the part of it that explicitly does the job you are expected to do. It’s not even the demands of others that cause this narrow focus; it’s often just that the error-handling parts are not as interesting to the person writing the code as getting a result. It would seem that many programmers just want to move those bits, munge that data, and show pictures of cats.

In point of fact we have a clear indication of the importance programmers put on the error-handling components of the code by this finding: “Error handlers with TODO or FIXME in the comment.” Personally, I prefer XXX, as it reminds me of my time in Amsterdam in the early 1990s, and unless you’re working in certain industries—industries that might also serve photos, and might still serve photos of cats—you’re unlikely to find XXX as a variable in the code.

We can look at the fact that the Java compiler forces programmers to catch all the unchecked exceptions in one of two ways. If we are charitable—and KV is the heart and soul of charity—we assume the Java language and compiler developers are simply helping programmers make fewer mistakes and make sure their code not only does what it is meant to do, but also acts appropriately when things go awry.

If we are less charitable, or perhaps more honest and realistic, we see this enforcement quite differently: as a naked attempt to control programmers and make them do what the language and compiler people thought was right at the time. “Programmers don’t do proper error handling. I know, we will MAKE them handle errors, or their programs won’t compile at all!” I believe this is said in the voice of an overbearing schoolteacher. “You will dot your i’s! You will catch all exceptions!” Except that unlike dotting an i, there are ways to skate around handling the exception that was meant to be handled. In a rush? Well then, just add a TODO or FIXME or XXX in the comments and move on. You’ll come back to it later … of course you will.

Both sides are a little bit wrong in this case. We can all point fingers at the person who leaves a trail of FIXMEs in the code, but who among us is without blame in that regard? We can also blame the pedants who thought that forcing every exception to be caught was doing us a favor. You can never discount the human element in programming. For everything you try to force on someone, there is something they will work to avoid if at all possible. Tool builders need to understand that the people who use their tools are often trying to get a very narrow job completed with a minimum amount of effort. Was it wrong to add the forced exception handling into the tool? Maybe and maybe not. In the hands of someone with the time and inclination to do the right thing, these errors are a welcome way of finding problems that they do have to handle.

Clearly, in the hands of a large percentage of programmers who work on some of the most complex systems yet devised, the feature is actually a nuisance, and it is likely time to rethink how this particular exception ought to be handled.

KV

q stamp of ACM Queue Related articles
on queue.acm.org

Productivity in Parallel Programming: A Decade of Progress
John T. Richards, et al.
Looking at the design and benefits of X10 http://queue.acm.org/detail.cfm?id=2682913

Going with the Flow
Peter de Jong
Workflow systems can provide value beyond automating business processes. http://queue.acm.org/detail.cfm?id=1122686

The Challenge of Cross-language Interoperability
David Chisnall
Interfacing between languages is increasingly important http://queue.acm.org/detail.cfm?id=2543971

Back to Top

Join the Discussion (0)

Become a Member or Sign In to Post a Comment

The Latest from CACM

Shape the Future of Computing

ACM encourages its members to take a direct hand in shaping the future of the association. There are more ways than ever to get involved.

Get Involved

Communications of the ACM (CACM) is now a fully Open Access publication.

By opening CACM to the world, we hope to increase engagement among the broader computer science community and encourage non-members to discover the rich resources ACM has to offer.

Learn More