Archive for April, 2009

Good Ideas Aren’t Always New

April 21st, 2009 16 comments

At QCon, Glenn Vanderburg, Michael Feathers and I (there may have been others, as I recall some ESB was involved…) were talking about Mike’s 10 Papers Every Programmer Should Read post (if you haven’t read it, please do so now, I’ll wait. No really, go on.) A lot of programmers aren’t particularly well read, a fact that Mike laments, and we kicked around some theories as to why that is – here’s mine.

Why don’t we look at our past? I believe it has to do with the natural (10 year or so) cycle of language dominance. People who’ve never programmed in language n-1 look at the syntax in a book from, say the early 90s and scoff; they complain that they don’t know that language thus they can’t read the book. I’ve heard more than a few developers dismiss Design Patterns because the code wasn’t in Java, of course Java is starting to wane – a couple of days ago I was researching a book and noticed it had a bunch of negative reviews because the example code was Java!

Many developers are essentially Blub programmers and they can’t imagine life in any other language. Further, they firmly believe that anything that existed before Blub is the modern day equivalent of the Pony Express, antiquated and useless. To some, all the good ideas are new, and we have nothing to learn by studying our past, I suspect many of these people have never seen the mother of all demos (though maybe we have something new to rival that now.) A typical Blub programmer assumes that any book that doesn’t use Blub isn’t worth his time and thus misses out on a wealth of learning. They almost willfully ignore the past which explains some of the reactions to Mike’s post.

I’ve been doing my best to read my way through 10 papers and this afternoon I came to Kent Beck and Ward Cunningham’s A Laboratory For Teaching
Object-Oriented Thinking
. Now, many people assume the whole agile software thing is just a few years old but here we see waaaaaaay back in 1989, a reference to YAGNI and a plug for teamwork:

We stress the importance of creating objects not to meet mythical future needs, but only under the demands of the moment. This ensures that a design contains only as much information as the designer has directly experienced, and avoids premature complexity. Working in teams helps here because a concerned designer can influence team members by suggesting scenarios aimed specifically at suspected weaknesses or omissions.

Great ideas then, a great ideas now. We’re a young industry, one that is further hamstrung by the belief that the language/technology/process du jour is all that plus a bag of chips. We’ve obviously made some huge strides, but in so many ways we’ve barely moved. But for the video quality, Engelbart’s demo could have just as easily been from 1988 or even 1998; sure, we’ve all got a mouse on our desk, but what about that 5 fingered keyboard? Not so much. Heck, every language is just trying to reinvent Lisp, and that’s more than 50 years old! Don’t be afraid of the past, those old guys knew a thing or two. We only hurt ourselves by ignoring the lessons they have to teach us.

Categories: Development, Rants, Software Tags:

Test Infecting the Legacy Organization

April 15th, 2009 13 comments

As Neal Ford explains, the NFJS Anthology series has been reborn as a monthly magazine and in the current edition, you can read my take on test infecting legacy organizations. I’ve been a proponent of the testing meme for most of my career but I’ve also spent much of that time convincing reluctant coworkers (and managers) that testing was in their best interest – the article takes my talk of the same name and puts it to paper. All NFJS attendees get a complimentary copy of of NFJS, the Magazine, but anyone is free to subscribe. Each month you’ll get an eclectic mix of articles written by NFJS speakers on topics they are passionate about; if you’d like to see a sample article, check out Jared Richardson‘s A Case for Continuous Integration [PDF]. Enjoy!

Categories: Agile, Articles, Software, Talks Tags:

Vendors are Risky Too

April 13th, 2009 4 comments

“We’re not a software company” is a common refrain these days; ever since Nicholas Carr’s “IT Doesn’t Matter,” it seems like more and more companies are bending over backwards to prove they don’t do IT. In the process, some consultancies have made a ton of money, often by “replacing” a given company’s IT department with the newly hired agency replete with people from….the company’s former IT staff. More than a few of my friends went to work one day as an employee of company X only to enter the office the next day via a contractor badge. As companies aped one another, we heard more and more about “core competencies” and how smart companies stuck to what they did best.

Implicit in this arrangement is a transfer of risk – and many (on both the business and IT side of the shop) equate vendors with risk free, or at least, if things go south we have a throat to throttle. While it can be empowering to scream at a vendor rep, that doesn’t mean you’ll get your problem solved – or that they’ll even be inclined to try. Vendor priorities mar or may not line up with yours and more often than not, that service contract entitles you to the C squad, not the A players they showed you during the courtship.

Make no bones about it, when you saddle up with a vendor, it’s a commitment, one you best enter into with your eyes open. Just like with your spouse, year two is rarely the same as the first date and having a phone number to call doesn’t mean you’ll get an answer you’ll like – or even an answer at all. Sure, you can open a problem ticket, but when will it be resolved? Don’t hold your breath, unless your CIO calls their CIO at least. Oh, and never assume the vendor’s developers write higher quality code than you do – some of the worst smelling balls of mud were slung by people working for “software companies.”

While we’re on the topic of software companies, don’t automatically think that the vendor is anymore of a “software company” than you are. It may *seem* like they’re in the same camp as Microsoft or Oracle, but take a look at their income statements – does “professionals services” make up a large percentage of the bottom line? Odds are it does, the software is the modern day equivalent of razors; they’ll darn near give it away (OK, if you call 7 figures “give”) so they can line up a nice fat services contract (mmm, smell the subscription fee!) In some cases, your odds of successfully implementing the project rapidly approach zero without a significant investment in contractors at $250 an hour and up. Nice work if you can get it.

Speaking of contractors, before you pull the trigger on that shiny box of vendor joy, take a look around the job boards to see if anyone is looking for developers with that skill set. Better yet, have your HR people look over recent job applicants to see how many boast time with your new love. Staffing models aren’t always at the forefront, but if you can’t hire people to twiddle the vendor bits, take a look around your cube farm and be sure you’ve got something people will want to train up on. Don’t be surprised when your techies aren’t thrilled by the notion of babysitting a piece of packaged software.

Once you’ve committed to a vendor, you live life on their schedule. Occasionally you might be able to nudge things but chances are you’ll be treated like the rest of the huddled masses. In some cases this might be just fine, but in others it can have a significant impact on your business. Found a critical bug? Odds are that won’t be fixed until the next release…sometime next year. You’re also stuck with their priorities and again, while you might have some influence here, more often then not, your pet feature isn’t so important to the decision makers at Vendor Co.

When you bring in a vendor, expect a platform play – and not all platforms are created equally. The excitement in Java land these days isn’t over Java the language, it’s Java the platform that gets the much deserved press for housing the likes of Clojure, JRuby, Scala, Groovy (hint, your developers would love to play with any of the preceding) and a host of others. While the Java platform (and it’s peer from Microsoft) offer a slew of choices, the vendor’s platform is probably designed like the Hotel California; once a company has invested time, effort and money, they will usually continue to throw good money after bad.

I’m not saying you should never purchase a vendor product – far from it. You shouldn’t write your own database server, you own app server, your own OR mapper or your own build framework (OK, maybe as a replacement to Maven, sure I can see that.) But when it comes to core competencies, the things that make your company special, that’s not something you should be too keen on farming out. As my friend Neal Ford is fond of saying, smart companies understand that IT is strategic.

When it is time to purchase some software, perform a true evaluation – and one that’s up to date. Just because the Foobaz team gave the product the thumbs up three years ago doesn’t mean they’d say the same thing today. Heck, that team might not have even considered the same criteria you are. Too often, we either run through a script that is oddly similar to the vendor’s demo or we try out a couple of hello world size examples; you’ve got to spend some quality time with a product to figure out when Dietzler’s law kicks in. Regardless of what you’re evaluating there are certain things you should pay extra special attention to:

  • The testing story. If a tool doesn’t have a good *automated* testing story, fail, or as we say in the project room, frog in a bag (FIAB for short).
  • Version control. Repeat after me: copying the files to the LAN isn’t version control. Life without source code management just isn’t worth living, if the tool doesn’t fit within something like Subversion or Git, your evaluation is over.
  • Can you diff the artifacts? I’m a fan of pictures, but show me the tool that can diff that fancy BPEL visualization. Oh and good luck with those massive XML files.

I’m sure you have other criteria too (Neal has a good list in part three of his SOA series) but these three are absolute deal breakers for me. Again, I’m not saying you should build everything, but when you do choose to purchase, be sure you know what you’re getting into. Can you live with the constraints? Are you comfortable with the tradeoffs? If so, I wish you all the best. But don’t blithely assume vendors aren’t risky.

Categories: Development, Rants, Software Tags: