Hacker Newsnew | past | comments | ask | show | jobs | submit | huyng's commentslogin

I glanced at this paper, here's my basic understanding:

Given an end goal, and a list of web-services, automatically compose the data "piping" required to reach the end goal. Do all of this using AI.

Can someone confirm this? If so that's pretty amazing and I will be giving this topic a deeper look.


It requires the OWL-S semantic annotations to be there in the first place unfortunately. Personally I have a hope that formal services will have some kind of buy-in to the semantic trends, way more than regular websites.


Thanks, I've always kind of grouped the business plan and this process into one thing. Makes sense to separate these two components.


Thanks for the response, can you clarify what USP means?


http://en.wikipedia.org/wiki/Unique_selling_proposition

haven't heard that term since advertising 201. what a bullshit class. i spent hundreds of hours creating hundreds of bullshit "thumbnails" for bullshit weight watchers campaigns.


Unique Selling Points?


No I don't think so ... I actually found this library as I was watching the GoogleIO Wave Keynote. When I saw what the wave team did with GWT, compiling Java => Javascript became a viable option in my mind. Before the demo, it had always been a mystery to me as to why anyone would want to use such an absurd amount of abstraction.


In Vaadin you do not have to compile Java to JavaScript. Instead you are using native Java on server-side. GWT is used only for compiling "client-side thinclient". This is only needed when you want to add more widets to Vaadin - if you are happy with the default widgets you do not have to worry about GWT/browsers/compilation.


Agreed, I definitely feel most comfortable with my Python on server-side, but the UI ability on this thing is beautiful. Too bad... I wish I was at JavaOne so I could ask these guys some questions regarding how easy it is to separate the two components.


You can ask all the questions on your mind on our forums (http://vaadin.com/forum/). You don't need to attend JavaOne to get the hard questions answered.


Vaadin keeps the application state on the server, but the presentation (HTML) is 100% generated on the client side. Changes in server-side state are serialized and sent to client in compact User Interface Definition Language (UIDL). This is obviously a more secure approach to writing web apps, since the state is safe on the server. So no pre-rendering of UI on the server..


Yeah. Every component has a server side and a client side. Client side has all the rendering stuff in it while the server side has the business logic. That way the functionality can't be hacked as the client side is just responsible of sending input to the server, where all the hard work is done.


According to this example: http://demo.vaadin.com/Coverflow/ It seems like you can add custom widgets easily. In this app they started using a Flash Widget for custom graphics.

I'm not a GWT developer, does anyone have any experience adding custom GWT/Vaadin widgets to your applications?


Book of Vaadin http://vaadin.com/book has a whole chapter on this. Also the Vaadin Eclipse Plugin makes getting started with adding new widgets trivial (just "Add New Widget" and it creates a widget stub for you)


I'm looking deeply into solid frameworks for developing rich cross platform apps. So far this looks promising as well as Capuccino. The killer feature for me is this table widget implementation http://demo.vaadin.com/sampler/#Components/Table%20%28Grid%2...

Here is the HelloWorld tutorial: http://vaadin.com/tutorial


This experimental search is very well done. Will definitely recognizes a problem a lot of us are having with trivial day-to-day stuff such as looking up documentation for a Django function or class. I know a lot of my time is spent looking up functions.

Because of this, I've been working a light-weight search-engine to sift through the Django source code:

http://www.djangocodesearch.com

Hopefully it'll help you find the code you're looking for, fast.

In regards to vertical search, I feel like the biggest barrier to competing with G/Y/M is the relative difficulty a user has to go through to reach the actual search page before entering in the information. If its more than a few key strokes away, then the user has a natural tendency to resort to the Big 3 engine which are situated conveniently in their search bar.

Adding a Firefox Keyword Search would definitely be a step in the right direction, but unfortunately the other browsers don't have that facility.


Josh Gourneau setup a MyCroft plugin which makes it pretty trivial to use Findjango 'in browser' for IE and FF, but not sure what the options are for WebKit or Opera.

I agree that exposure is a huge issue there, as we're not likely to sign and distribution deals in the near future.


This originally started out as my own tool, but I thought you guys might be able to benefit from it. I'm always forgetting how certain functions such as "direct_to_template" are called , so I end up digging through source code to find out.

This lets you find code quickly. No need to dig through documentation to reference simple functions or class definitions.


Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: