Cay Horstmann

Thank you! I've been wondering why there aren't more nuanced ORM perspectives in the blogosphere. Yes, OR mapping is a hard problem, and using an OR mapper without some understanding what it does under the hood isn't going to be wonderful. Yes, if performance is paramount, someone out there can write better SQL or NoSQL than the ORM could. But there are an awful lot of use cases where reasonably savvy people use an OR mapper to save themselves a bunch of menial labor, and performance is just fine. It's a tool, not a silver bullet.

Scala may not be a scripting language, but it is a great language for writing scripts. You can do a lot with small snippets of code, and you have the REPL to experiment, just like a traditional scripting language such as Python. The best part is that you don't have to learn two languages, one for scripting and one for heavy lifting.

I am not sure that the "everyone is beating on poor Oracle" meme captures the nuances of this issue.

When you look at the release notes at http://www.oracle.com/technetwork/java/javase/jdk7-relnotes-418459.html#knownissues, one issue does stand out over the usual "grief with Java plugin" or "grief with CJK input" or "grief with weird X11 issues", namely that an optimization bug can sometimes silently deliver the wrong result.

People don't like to get the wrong result, even if the probability is less than them getting struck by lightning. We all know that because it's happened before--remember the Pentium bug?

So, if Oracle had said "Whoa, we just found out this issue, and we'll fix it immediately, but it'll take ten days to re-run all the acceptance tests", nobody would have batted an eye. Or if they had said "Look, we've got this problem, but you've got to ship sometime, and here is how you work around this vexing issue", people would have been ok. Their problem was to say nothing at all. If you say nothing at all, you open the floodgates to "Don't use Java 7 if you use loops", "Java 7 unsafe at any speed", "Sloppy work a Oracle", and all the other hyperbole. That's a lesson well learned for anyone who needs to make a similar decision.

@RawThinkTank: Actually, the _ is a function parameter. _ % 2 == 0 is the same as the function x => x % 2 == 0. If a parameter only occurs once, you can use an _. For example, _ + 1 is the "increment" function, and _ > _ is the function with two arguments that compares them--you can pass this to a sorting method to get sorting in reverse order. ,

Well, I am glad someone is willing to go after the sacred cows. I used to obsess over "aggregation vs. association", but when UML sprouted more metamodels than you can shake a stick at, I realized that it was a lost cause. I am all for standardized class and sequence diagrams, but beyond that, I can't think of any part of UML that has caught on. One reason is...just try using the damn things. There is a lot of complexity that basically says you aren't allowed to draw what you want to. I know it must be my fault because I can't comprehend the metamodel :-(

"I would love to see some good examples of operator overloading, if you have any to share... "

Check out the Scala combinator parser library. It would be really tedious to use with the traditional method invocation syntax.

Of course, that's Scala, not Java, where you can define your own operators. In Java, there is a fixed set of operators, and overloading is much less interesting. Maybe it really is time to leave Java alone.

"I would love to see some good examples of operator overloading, if you have any to share... "

Check out the Scala combinator parser library. It would be really tedious to use with the traditional method invocation syntax.

Of course, that's Scala, not Java, where you can define your own operators. In Java, there is a fixed set of operators, and overloading is much less interesting. Maybe it really is time to leave Java alone.

Odd how the code for turning the book pages is longer than the code for showing the rotation of the planets

Here are two of my favorites that haven't been mentioned yet:

• https://woodstock.dev.java.net/
• https://xhtmlrenderer.dev.java.net/

As I understood it, the point of the NYT article wasn't so much that many citizens are dumb--which surely has been the case throughout many cultures, throughout much of history--but that being dumb is socially acceptable in America, and being smart marks you as a freak.

I don't think that you see that kind of anti-intellectualism very much among software developers. Most successful developers are proud of their knowledge, seek to enhance it, and respect those who are smarter than they are. After all, you don't get very far in this business by being dumb.

I think the more interesting question is whether other cultures value knowledge--and knowledge professionals such as software developers--more than the U.S. does. Maybe the overall culture has an influence on the quality and quantity of software developers? Many of the smartest developers I work with grew up in China, India, and Eastern Europe.

I grew up in Germany, and while I can't say that I got any social brownie points for being good in math and science in high school, it didn't stigmatize me either. My wife grew up in Vietnam, and she says that in her high school it was actually a social negative to be bad in math--what is fun about dating someone dumb? Here, in the U.S. of A., my children would be too ashamed to tell their classmates the truth--that daddy got a degree in math. They'll say that I run numbers on the waterfront.

Is this a recent development? In the 1950s and 60s, American science was the envy of the world, and we still benefit from the infrastructure investments of those times. But there is precious little investment these days, perhaps because the social norms have changed. That should worry us because it affects our business.

Cheers,

Cay

I find it interesting that you say that people pick up JSF the fastest. Maybe it isn't as complex as its detractors make it out to be :-)

I find that my students pick up JPA very quickly as well, so Seam sounds like a winner for gluing the two together.

Seam doesn't need Hibernate persistence, but it is a bit of an effort to strip out the unnecessary pieces when you deploy on an app server other than JBoss. I think it would be good if the Seam packaging was more vendor-neutral.