Comments on: Choosing a web development framework/toolkit

By: Matt

Matt — Tue, 31 Aug 2010 00:06:00 +0000

Great post, I couldn’t agree more about domain modeling. I discovered Grails recently and with its Hibernate support I was hopeful that the framework would be a complete solution, but alas it’s not nearly flexible enough in terms of separating persistence from your domain and creating a rich domain model.

By: Simon Palmer

Simon Palmer — Sat, 23 Jan 2010 08:16:28 +0000

I liked reading this post and the subsequent thread and I think I agree with much of what you said, thanks for taking the time to share your thoughts publicly. How do you feel about your choice of Grails 6 or so months on? Has it continued to deliver benefit to you? Are there any obvious problems? Are you busy coding round any one particular aspect of the framework? I ask because I just took the decision to go Grails for a new project and am getting increasingly scared that it is unlikely to perform or scale. I have a lot of database design experience and I am suspending my disbelief that hibernate in its default GORM incarnation is going to get the queries right. I deliberately want an almost join-less database model and am already finding myself using composition to keep the code neat and the domain model sane. This does seem to cut against the Grails convention as you mentioned and I hate having to make my domain model and persistence model identical. Rather than that being an acceptable convention it feels like a naive restriction imposed by a Java framework team who may never have had to build really scalable database applications – but maybe that’s being a bit harsh. Incidentally I recently spent some time coding C for embedded devices after a long time in Java and Web frameworks. The key thing there is that everything fits inside 64k. Coming back to the incredibly wasteful world of normal programming makes me wonder when the framework fad will shed its obesity and free up all those CPU cycles that it spends doing stuff which made the programmer’s life easier.

By: Simon Palmer

Simon Palmer — Sat, 23 Jan 2010 08:16:00 +0000

By: web 2.0 development company

web 2.0 development company — Thu, 12 Nov 2009 07:08:42 +0000

there are many points things needed for to start web development . . . as you written you gotta done at all above in the article. . you just go ahead & start the journey of web development . .

By: web 2.0 development company

web 2.0 development company — Thu, 12 Nov 2009 07:08:00 +0000

there are many points things needed for to start web development . . . as you written you gotta done at all above in the article. . you just go ahead & start the journey of web development . .

By: Kevin

Kevin — Tue, 11 Aug 2009 20:18:19 +0000

I hear what you are saying Ilya. I don’t know that I’d want to hire someone with a little play time in multiple languages for my java based system over someone that has worked solidly in java but doesn’t have any experience in the others. It seems to me an expert in the language you are using is a far better choice than someone that knows it decently but also brings to the table a couple other languages which you don’t use. Oddly enough, in my early 20′s, when I had lots of energy and no kids, I loved learning new frameworks and languages. But, today it is a lot different… we have huge resources available to us via the internet, and I’ve barely tapped the Java language to know it’s full potential, let alone trying to spend time that I could (and should) be using on my day job learning a new language or two that I most likely won’t be able to use. By toy language I mean they are barely an itch in the industry compared to Java or .NET, and that so far most people I see using them often are the ones that didn’t care for Java or are tired of Java for one reason or another and want something new. Despite many sites popping up with php, ruby and python, they are still minuscule compared to the jobs available and sites built on java, or even .NET. I look at a couple of other things too. For example, iPhone’s language is objective C. Androids is Java. These two potentially huge platforms (iPhone already is and Android is fast on its way) chose these instead of languages like groovy, ruby, etc. I can’t help but ask, if Ruby, Python, Scala and these others are so much easier to use, faster to develop with, why not choose a language like this to write small phone apps over the larger robust languages? Maybe this isn’t a fair argument, but it does make me ask why the simpler faster languages were not chosen? See, back when there were a couple main languages, like assembly, pascal, c/c++ and java, they were all used quite in abundance…even Basic. But now, we have a crap load of “the next language to replace java” springing up, most of which are scripted languages not even compiling languages, and the few that use them rave about them and some how get a small following of people that jump aboard and hype it up as if it’s the next big thing. I see tons of blog posts about how scala will replace Java, how groovy is going to replace java, etc. Yet, nothing has come close. Even Ruby and PHP aren’t even close. They stand the most chance to make a fraction of a dent, but won’t ever come close. I honestly kind of get tired of seeing all the rants about having to learn all these languages. I have enough to do with Java and one job to keep me busy for years to come, and I truly love the language. I spent the time to learn a lot of it’s inner workings, to think about having to go do that all over again because a fraction of developers who are tired of Java for one reason or another (most of which have arguments that really aren’t that valid) say so I don’t know.. it just doesn’t seem like it’s necessary. Now, let’s say there was no ruby/rails, php, python, etc and some newer language, pure OO, cleaner syntax, etc came out, it might be worth looking in to if it really offered something better. But I haven’t seen any argument yet that indicates any of these languages are better or offer anything more than what Java does already. The biggest argument is speed of development. Hmm.. copy/paste of tried/tested code works for me. Java is a well understood language, with tons and tons of money and sharp minds behind it, using it. I don’t see anything Ruby, PHP or any other language offers me that is better than what Java does. It’s almost like the “If it ain’t broke, don’t try to fix it” analogy. Java comes on every OS out there now, Ruby and PHP do not. I know that anything I write in Java will almost always run on any of the OSes out there out of the box (give or take the few that aren’t updated to 1.6 yet.. thankfully OSX finally is.. but 1.5 is still there), I can’t say the same for any other language out there except maybe C. I could probably go on all day. You did make a good point.. to me php and ruby are very similar to perl and I only dabbled in perl. Like I said in my first post, I find the language syntax to be confusing and hard to read. Maybe it takes time.. but every developer blog I read about these languages raves about how fast and easy it is to develop in, and I don’t consider myself a beginner developer but after looking at source and reading a book on Ruby, I still find it hard to work with and cryptic to read. Maybe my mind is stuck in the Basic/C/Pascal/Java like code style, but they make sense to me. These others ones are strange to look at and remind me highly of shell scripting and perl, neither of which I care for. lol.

By: Kevin

Kevin — Tue, 11 Aug 2009 20:18:00 +0000

By: Ilya Sterin

Ilya Sterin — Tue, 11 Aug 2009 17:28:48 +0000

Kevin, I have dabbled in ruby/python frameworks for a bit before forming my opinions. I wouldn’t consider myself an expert in them and wouldn’t necessarily say that my mind wouldn’t change if I’ve used one of them for quite a while. Ruby or Python didn’t seem foreign to me, hey, I was a Perl guy before Java To say that these languages are toys is pretty uninformed to be honest, they are true proven languages. Perl’s probably running as many if not more production critical systems than java, it’s just not as hyped any more. Amazon’s store front is mostly written in perl. Python powers lots of google’s infrastructure and PHP powers yahoo, so not sure what you mean by toy languages. Now with that said, most good programmers learn different languages even if their day job doesn’t allow them to use it. New ideas, ways of thinking, etc… are spawned through learning different languages and frameworks. I’d much rather hire someone who’s worked in a heterogenous environment, understands the concepts and is not proficient in a particular language we’re looking for, than to get someone who only programs in Java and uses it’s echosystem. There is no doubt that java and it’s libraries are probably the most stable and sufficient frameworks for developing enterprise apps, but you also must look at other factors. Thankfully with the advent of other languages running on the JVM, you now have a choice. I wrote up a piece a few weeks ago on Groovy SQL builders for example. I could have also done it in Java, but what a pain it would have been. Now, to say that we want to save a few lines (actually many lines) of code, is true. But I don’t see anything wrong with it. That’s the nature of computer science, building abstractions, otherwise we’d be writing code in assembly and telling java developers how dare they want to save a few lines of code. There are times when it’s important to understand the lower level details, there are times when it’s not as important. Now, even if I do understand it, I wouldn’t say I’d really want to say parse raw HTTP packets, even though I’ve rather familiar with them. I’d rather the framework take care of that for me. The same is true for other cross cutting concerns. If we can get an abstraction to take care of it for us, more power to the abstraction. Yes, there are side cases when you have to get dirty into the implementation details, but with good abstractions, these should be far and between. There are of course “leaky abstractions”, which I still think are beneficial, they just expose the underlying problems with what we’re trying to abstract. I love java, i’ve programmed java for the last 10 years, but with that said Java is not the end all of software development. There are better suitors for particular problems. Like Erlang or Scala for distributed and multi threaded programming (have you even tried multi-threaded programing with java’s thread primitives? If you’ve ever developed any serious multithreaded app in java, you’ll know that it’s super hard, it’s non-deterministic, and because of that it’s bug prone even when you tested the heck out of it). You also have Groovy, which brings some level of dynamic programming to Java. I don’t particularly love groovy, but I like it’s interop with java. Also, Clojure looks awesome for functional abstractions and multi-core concurrency through STM.

By: Ilya Sterin

Ilya Sterin — Tue, 11 Aug 2009 17:28:00 +0000

By: Kevin

Kevin — Tue, 11 Aug 2009 16:43:35 +0000

For the OP.. I am curious. did you know Ruby and grails before taking it on after you felt Spring MVC was too cumbersome? I’ve dabbled in Ruby a wee bit and find it’s language syntax quite cryptic. Haveing done OO Pascal, C/C++ and Java, they feel like languages to me. Ruby, Perl, PHP, those scripting languages feel out of place to me. I know if I spent the time I could learn them, but why? So far I haven’t found any reason to switch because Java is far more robust for much more than what any one of these languages offers. To be fair, JRuby and those that run on the JVM have the JDK stack available, but I don’t see the point in using a new language just for the sake of learning something new. They feel like toys to me and while they may offer some of those that just don’t like the Java language some new way of doing the same thing, at least to me, Java is a much more powerful platform than any other language out there. I am also amazed at how many Ruby developers say how much faster it is? Forgive me for using copy/paste, but I can get a full scale J2EE app complete with EJB, mysql integration, MVC, separation of tiers, etc in a very short time, couple hours tops if not less. Then I just start adding my code for specific functionality. I am failing to understand the developers, especially those coming from Java that rave about how fast it is to develop in Ruby. Now on to your topic specifically. You mention REST, MVC, and many other things. It sounds like you want the end all framework and to this day I don’t think any provides it all, although combining Spring + Hibernate or EJB3, and a couple other frameworks sounds like it would do what you want. There is no single framework that does what you want, and I’d guess there never will be with so many solid frameworks already available. Short of you developing your own from scratch, I can’t imagine someone skipping out on all the good frameworks now available to start yet another one. Most can do in some way what you ask for, and with a little bit of inter connectivity, you generally can get anything working within Spring. Mind you, I am no Spring aficionado, I’ve used it a little bit, that’s all so I am not trying to plug it. So.. one thing I find really good is using Jersey for it’s REST capabilities and using it from AJAX directly to bypass the need for a pure MVC framework. It allows me to offer 3rd parties a solid REST api, and I can use it as well from my own web UI with AJAX. It does NOT do the form auto population into objects, validations, etc. But I don’t often find that too hard to put in anyway. Even so, I could use another framework or sub projects out there to do most of this anyway. I haven’t checked recently, but I’d guess there are some smaller validation libraries and form to object population libraries out there. What I find most puzzling regarding your post and many other developers is often the complexity that they seem to go thru to avoid doing work. The configs and such to get all these parts working to me means more moving targets for problems at deployment, as well as I often find I spend more time writing xml config code to get thing to work together than I do writing code. These days, using the Jersey REST library with EJB3 I find I can quite easily handle most everything right in code without any sort of configuration issues. I have the jersey config in my web.xml, the rest is in code and works great. No multiple configs to worry about, less dependencies on various frameworks and what version they are and what they depend on and what I am missing on my classpath. It seems as we’ve all grown into these large robust frameworks, we’ve add a lot of extra complexity to make them work just to try to save a few lines of code here and there. Don’t get me wrong..I think they are great, but I think they add complexity that is often not needed. I’ve seen numerous hours, days sometimes with a simple deployment due to various dependencies and developers not knowing how that all works. It’s often too easy to read some tutorial online about how to get this to work with that, and that is all they do.. they don’t understand the inner workings and it causes more problems than it solves. It’s much like developers that don’t know much about how the classpath works, how classloaders work, how memory management works in Java. Anyway.. just my thoughts on this.