Skip to main content

Which of these new Java language features are you most looking forward to using?

Generic types
37% (444 votes)
Metadata
15% (172 votes)
Enhanced for
15% (173 votes)
Enumerated types
7% (81 votes)
Static Import
7% (81 votes)
Other
20% (234 votes)
Total votes: 1185

Comments

How about: none of the above?

Improvements to JVM? YES Changes to language to make it harder to read and maintain? Why God why???

Covariance!

I'm really looking forward to covariant return types... it's kind of part of generic binaries I suppose; I think it's a great addition to the self-documenting nature of interfaces.

The important things

Java already is the best language for programming. Therefore now the really important things for improvements are - Fast application startup - Effective memory management - Easy VM installation thank you

The important things

I would have to agree with your points, however, having templates for Collections (while increasing the difficulty of readability) helps a lot.

method variable number of arguments

The methods with a variable number of arguments seems to me [helped me in my projects very much lately] the most important new thing in java

Broken poll

You can vote multiple times and each one is counted. The polling software is broken and results are not valid. Period.

Not really a Broken poll

Thanks for your thoughts - I have taught stats many times and agree that this is not a "valid" poll. However, here's a bit of what we were thinking.

There is a wide range of polling from identifying a random sample and hiring an independent polling organization to select, phrase, and ask the questions to those bumper stickers on the back of trucks that give a phone number to call if you want to comment on their driving. In that spectrum, we are a lot closer to the truck end. Our poll is for a self-selected population that can, if they choose, vote more than once.

The next step up the chain could have been to require login and to enforce one vote per java.net login. We decided that what you lose is some of the ease of use. If someone sees our poll question and wants to register an opinion they are just a click or two away from doing so. What you would gain by this policy is that only java.net members could vote and they could only vote once per login name. Of course, some members have more than one login and the polls would still be biased because those voting are those that feel strongly enough to want to vote. In other words, the results would not be that much more statistically valid.

So we've decided, like many of the sites we looked at, to make it easy for people to participate. We hope that people won't abuse the poll by voting multiple times but we expect that people referring to results will understand the mechanism used for casting votes and interpret the results accordingly.

If someone feels so strongly that varargs is the most important feature coming in Tiger that they vote more than once for it, you are right that their vote is recorded more than once. The polling software is not broken. It works exactly as it is intended to work. Whenever you read a statistic you need to understand who gathered it and how.

Thanks, Daniel

Yes, it really is a Broken poll

Thanks for your note. Many polling software packages will use cookies to track whether you've voted or not. If you've voted, it shows the results in place of the poll itself. It can be gotten around, but it discourages multiple votes. Secondly, you can also allow only one vote per IP address. Another method is to store a session variable flagging that the user has voted in that poll and should not be allowed to again during that session. Alternately, only 1 vote per IP address per 10 minute span prevents repeat votes. They can still vote multiple times if they "feel strongly", but it takes more time and prevents 100 votes in 3 minutes. Using one or all of the above does not require any logins and at least the first method will a) prevent accidental or "casual" multi-voting, and b) provide more value to the java.net visitor as they don't have to do an extra click to see the live results. I see where you're coming from regarding how the poll was set up, but my guess is most people will not attempt to understand it's reasoning and dynamics. Most people will vote, peek at the results, and move on. Most people do NOT try to understand "who gathered it and how" and will make "default" assumptions. They will generally assume it is reflective of the community at large rather than the "strong feelings" of a particular member who may or may not be representative of the population. I base my statements on a long history of polls on the net, a majority of which do NOT allow multi-voting. Most people will assume this one is like the others. I hope you consider one or more methods described above to discourage multiple votes. If not, I hope you will find a method to count and display the number of voters in addition to the number of votes for each item. If we don't know how many people voted, it's useless in my opinion. For all I know, only 3 people voted, not 300. I don't know how big a grain of salt to take with the poll, if you know what I mean. Thanks again for your response, though. But it's still broken. :-)

Yes, it really is a Broken poll

Official voting on java.net (elections, etc) will be done is a manner that prevents duplication.

Your points are valid regarding the quantitative accuracy of the poll. Luckily, the spirit of the poll is to be informative and not something that major decisions will be based upon.

As you mention, there really aren't any non-intrusive ways to completely prevent multiple votes by a single individual:

  • Cookies can prevent a single instance of a browser on a single computer from submitting multiple responses. It will not necessarily recognize me when I come in from a different browser or computer.
  • Requiring membership, login, or authentication is not in the spirit of what we are trying to do and is not conducive capturing the widest audience.
  • IP specific approaches also identify machines and not voters. So, public computers in libraries, airports, and schools can only vote once. DHCP and NAT present a whole other set of glitches.

You are correct in that there is value and merit to having statically accurate poll results and we would like to get closer to that in the future. I am just glad that we finally have polling on java.net!

For now, it will have to be like a Florida election.....you can "vote soon and vote often". ;)

Hopefully, that won't get abused.

Broken poll

If you've ever taken a statistics class you would know that a self selecting survey such as this is never valid. The fact that you can vote multiple times really doesn't make it any worse.

Broken poll

True if the intent was a survey. This is a poll, which is a bit different. But a single-vote poll is certainly more valid and interesting than a ballot-stuffing poll.

Broken poll

Sure its valid. My vote(s), 200 of them, just count for more than your single vote. And, I get to make that decision! Cool! ;-)

Broken poll

If you read my note carefully, you'd realize that I voted more than once. The question is, how many times did I vote? Do you know for sure that I didn't vote 400 times, making the ratio 2:1?

more interested in the library

concurrent utils in particular, plus improvements to swing. i'd actually like to know more about what other library additions are in there, besides the reworking of current libraries (collections, reflection) due to the language changes (generics, autoboxing, varargs). and what other existing libraries have changed as a result of the new features?