[Haskell-cafe] GSoC Proposal: A 3d visualization of code (packages)

Tillmann Vogt Tillmann.Vogt at rwth-aachen.de
Mon Apr 4 11:36:27 CEST 2011


Am 04.04.2011 11:06, schrieb Vo Minh Thu:
> 2011/4/4 Tillmann Vogt<Tillmann.Vogt at rwth-aachen.de>:
>> Dear Haskell Programmers,
>>
>> To get some feedback on my proposal here is posting that explains it more
>> detailed:
>> http://tillmannvogt.wordpress.com/2011/03/30/gsoc/
>>
>> If this succeeds it could become the poster child application for WebGL and
>> there will be an immediate value to all Haskell programmers: Get an overview
>> what is happening. In the long term there is the possibility to have a very
>> powerful UI-library and to be able to combine visualizations with Code.
>>
>> The milestone plan is just to show that it should be doable. The part that
>> is not so clear how to do is the integration with hackage2. It could even be
>> a separate site like haskellers.com that links to various resources. Should
>> a WebGL engine be used (then it should be easy to load a collada file) or
>> can WebGL be integrated in a haskell web framework?
> Hi,
>
> I'm happy to see someone interested in hacking 3D projects in Haskell.
> But I'm failing at imagining your package visualization. Would it be
> possible for you to craft some image with some real data from hackage,
> and show how it would be more useful than, say, a normal web page?
>
> Let me quote the three points from your blog:
>
>      1. The visualization will let programmers quicker find a library
> they need (see changes, etc. ), because the visualization algorithm is
> like reversed pattern recognition to produce most recognizable shapes.
>      2. The code size, popularity, dependencies, whatever the community
> demands can be integrated or left away
>      3. A general way to extend or change the visualization
>
> I really think a more concrete (manually crafted) image would help a
> lot seeing what you mean by 1.
  It would look a lot like the second picture with the ~300 blocks. I 
would rather like to generate the picture automatically than to spend 
some days with sketchup.
> I'm wondering how, in 1., it makes it quicker to find what I need if
> it is not clear what kind of information it will provide. I say it is
> not clear as point 2 is asking the question of what data is available
> in the visualization.
The quickest way to find a library is to know the name and search it in 
google or with Ctr+F on the website. But if you have forgotten the name 
but roughly know the size or to what area it belonged (i.e. graphics) 
you could find it in the overview mode. If you search for a tall green 
building (assigning code size to building size) one can find it in an 
instant on the second picture. Try to find a library in a listing of 300 
names.

> Point 3 doesn't seem realistic to me. It sounds like a 3d
> visualization framework is on the way, which is a big undertaking.
Graph drawing is hard, thats why I wouldn't try to reprogram Graphviz in 
3 months, but softwarevisualization is in my impression not so hard. It 
like the overly quoted Lindemayer-systems that you can achieve a lot 
with very few commands. A framework sounds a little big. Currently I 
just want to see some images like the second. But I have approached the 
generation part as formally as possible to do it right.

> Cheers,
> Thu
>
>




More information about the Haskell-Cafe mailing list