It's a polyglot world after all - Lift adding support for JRuby | Hacker News Reader

It's a polyglot world after all - Lift adding support for JRuby

groups.google.com

50 pointsicey16 comments

Lift is hands-down the best web framework available today...

Can someone a little less involved please comment on the validity of this and the claims that follow it? Is it something I should be checking out?

I've tried using it a few months ago. It's quite hard actually. The docs were poor, understanding the concepts was not easy and in a lot of places it abuses the Scala syntax:

LiftRules.rewrite.append { case RewriteRequest(ParsePath(List("shopInfo", "view", id), _, _, _), _, _) => RewriteResponse("shopInfo" :: "view" :: Nil, Map("id" -> id))

My opinion is that I don't like. At all! There are better and simpler ways to do things, even the real-time chats they brag about (socket.io, anyone?).

One of the things I asked dpp for was a real introduction. He's very excited to talk about Lift's comet stuff (and it's easily the best comet impl you can ask for), but in the process he sort of leaves people who just wanted to get "hello world" up on the browser behind.

He's doing a round of lift classes, so hopefully some of that material will migrate to the website and make lift more approachable.

The video from the first Lift training is available here: http://vimeo.com/groups/75631/videos (skip the first 14 mins). It covers what you need to get started with Lift.

That's great, and the content seems good, but I still feel like a 2-hour video is sort of a glib response to "there is no good introduction to the framework!" :)

It is pretty good. Is it the best? I couldn't say. Some misc. good points:

- It runs on the JVM, so you get performance and libraries, but you don't need to go through all the Java BS to get something working. (Seriously, performance is overlooked. If performance is not an issue, why did Facebook write HipHop?)

- Comet support (via Actors) is great and really easy to use. Ajax as well. Writing interactive website is very easy.

- The "no plumbing" approach is pretty good. You don't have to setup routes of anything (mostly). You just embed callbacks directly in the page in a lot of cases and things just work.

- Templates are nice.

The bad points:

- The ORM situation isn't great.

- The documentation is terrible (but improving).

Overall Lift is certainly worth looking at for some good ideas. Having a concise language and simple toolkit on the JVM is great.

> If performance is not an issue, why did Facebook write HipHop?

Energy. Facebook has thousands of servers running PHP code. HipHop allows them to have the same performance with half the servers[1]. 50% less servers at that scale equates to non-trivial energy (and cost) savings.

[1] http://developers.facebook.com/blog/post/358

It's not nearly enough of an improvement over Rails or Django to displace them. I expect it to do much better over the next few years than any other functional programming framework but I don't expect it to escape that niche.

Pollak has realized, as I begrudgingly did, that Scala, like all the other type-inferencing static FP langs, is just too hard for the programming mainstream. Once you've decided to go polyglot you might as well just use a boring but safe core language like Java or C++ and build as much on top of that as you can in your dynamic language of choice.

If you have a small team of above-average developers and truly commit yourself to learning "the lift way", including all the power of scala, I believe Lift will get you very far.

https://groups.google.com/forum/#!topic/liftweb/ViklDFsBtPo quote: "Our experience with Lift was the best possible, and an important part of that was having Scala available within the framework. No other language could have allowed us to do what we did within such a short time frame, Ruby included."

Very cool! When I first saw this I thought it was going to be about Lift developers being able to use ruby code- there are a lot of Ruby packages out there that they could benefit from using.

Just from looking at Scala, it seems like a superior language, and from looking at Lift it seems like a well done web framework. However, a lot of the benefit of Rails comes from the ecosystem where there is a plugin for your every need. I doubt that Lift can offer enough benefits to offset that.

Maybe they can use this for their pattern matching: https://github.com/waves/functor

Interesting that you say Lift developers could benefit from Ruby packages; generally a big pro for Lift/Scala is access the the multitude of Java packages. JRuby, Scala, Clojure and other JVM languages all benefit from that ecosystem.

As someone who uses Lift (in addition to other frameworks) I'm a bit surprised by this decision. Lift is at the state where it is just getting the features and docs to be useful to a wider audience. Scala adoption is rising. It seems David Pollak is going for the corporate market and I don't think this is the right focus at this stage of Lift's development.

Is Ruby making inroads in the corporate market?

I think so. Thoughtworks, who are a straight-up corporate consultancy, have done Ruby since 2006. Ruby is certainly further along the adoption curve than Scala, so perhaps DPP is trying to leapfrog over the adoption of Scala by adding JRuby support. However, IMO Scala is not what is holding Lift back -- it's lack of decent documentation. Just read the other comments on this post...

I've been doing Rails work at ThoughtWorks for more than three years now. Judging from the projects I've been on, Ruby has achieved a fair amount of penetration with the Fortune 500 crowd. Whether that's expanding, topped off, or contracting, I can't judge.

My personal impression, backed by no real data, is that Ruby adoption has tapered off somewhat. Python and PHP web frameworks have absorbed the innovations of Rails and the subjective advantages of Ruby as a language over either aren't enough to close the gap.

I wouldn't say that Ruby or Rails are dying but it also doesn't seem to me that they've achieved escape velocity either.

Reply on news.ycombinator.com