Poll: Release Web Application Framework Or Not?

1 points by volida ↗ HN
We are trying to decide whether or not to make public our web application framework leveraging the power of J2EE.

Would you care for a new framework? What would you expect from a new framework in order to deploy a real service based on it?

7 comments

[ 2.0 ms ] story [ 29.3 ms ] thread
Like Everything else- It depends..

Does it fulfill a new and unique roll or is it mostly a similar implementation as existing frameworks?

For example- GWT and Cappuccino bring something new and interesting to the space- Program in either Java, or an Obj-C like language.. There are a lot of decent ways to write web code using Java- Spring is very popular for plumbing..

So the question comes down to.. What do you offer that no one else does?

The second question is, How much effort do you want to put into supporting this? It sounds like you've developed an application, and then pulled the framework out from it.. (Not a bad policy at all. Look at 37Signals and Rails)

If it's not your core product, and you'd be giving it away, how much time and effort do you want to invest in writing documentation, test cases, examples, etc?

How much time do you want to spend ensuring that a stable version remains consistent for years, rather than being able to upgrade the framework when necessary?

It could certainly gain publicity and popularity if it 'took off', but that goes back to the first two questions.

I think your poll questions are the wrong ones, to be honest.. How can anyone answer if they'd use it, before they know the unique selling points, and what it brings to a table.

How long is a piece of string?

On the other hand,

Actually, it has been stable for the last 4 years and used to deliver a number of applications.

However, these are good points. And defenetely it needs effort and time.

The questions were mostly targeted to measure people willingness to try a new one.

Personally I would be interested in seeing something new. Well, we do have Maven, Jetty, Spring and Hibernate to do agile development in Java (though - it's not even close to RoR's productivity :-) -- so why not to fill this gap? And beside that one, there are a lot of empty spots in Java-around.
I'm surprised that "performance graphs" gets any votes, at all. Even Rails (and Django and Catalyst and Cake) is "fast enough" for the vast majority of web applications, and if you can't build something as fast or faster in Java, you need to be going back to school rather than thinking about inflicting your code on the world.

Since we know nothing about your framework, I voted "who cares?" (because I just can't take time to care about every new "framework" that comes along, even in languages I like), but if there were a, "Huh? A Java framework in 2008? How does that even happen? I didn't even know people still built websites with Java." option, I probably would have voted that way instead.

"performance graphs"

I haven't personally tested RoR but from a quick recall of some tests I've seen for RoR on the Web, I think our performs several times faster, and there is always room for optimization.

The point isn't if you can or not build something faster than XYZ but rather showcase that you did.

My point was that performance is so far down the list of things I would care about in a framework that anyone who cares enough to put it as one of a very few limited options probably is not thinking about the problem domain in the same way I am. My surprise is that people agree with you that if you're listing a handful of possibilities, performance over other frameworks is a consideration. I'm just finding the whole thing bizarre, that's all.

Besides that, performance of widely varying frameworks written in widely varying languages usually can't be compared in meaningful ways. It's either a micro-benchmark of a few very specific test cases (and thus, not all that useful for predicting real world performance), or it is a "real world" example, but implemented with possibly wildly divergent quality and attention to detail--if you are a Java master and you build a test app, and then do a 1-to-1 port to Python, nobody would be surprised that the Java app outperforms the Python one.

Anyway, my point is that I simply don't understand how performance could be a meaningful piece of information in this context or how being "fast" could possibly make me want to use the framework over existing frameworks. Writing it in C would be "fast"...but I wouldn't use it.

Actually I would expect that people who already use RoR for example wouldn't really have performance a priority.

People using Java I think would be more sensitive to performance issues.

I agree that speed is not the top in the list for a framework, but having it in mind is important.