roassal 1 -> 2 todos

Hi,

Just an update: Milton and I worked on the treemap and its application in FAMIX system nesting map, and we now reached a working version. Thanks, Milton!

It is similar to the Roassal1 TreeMapBuilder implementation but it seems that it is not quite so quadratic

Nevertheless, the API makes the opening about twice faster and the Roassal2 engine provides for a much smoother to manipulation (drag, zoom).

<annotated-classes-argo34.png>

Cheers,

Doru

On Thu, Oct 30, 2014 at 11:33 PM, milton mamani <[hidden email]> wrote:

Hi Tudor

please try this in the last version of Roassal2

| b |

b := RTTreeMapBuilder new.

b

extent: 400@400;

weight: [:n | n ].

b shape if: [ :n | n even ] fillColor: Color blue;

if: [ :n | n odd ] fillColor: Color lightGray.

b from: (1 to: 10)

using: [:n | n < 5 ifTrue: [ { n+1. n+2 } ] ifFalse: [ {  } ] ].

b open.

Cheers,

Milton

2014-10-30 19:11 GMT-03:00 Tudor Girba <[hidden email]>:

Excellent job, Milton!

I reviewed the code and it looks good. You just made the code twice faster :).

Now, we need to refine a bit the API:

- We should have a variation of explore:using: that allows a collection as a starting point. In this case, we should create a fake root object with the collection.

- I would more prefer something like buildFrom:using: or from:using:. Then we can have buildFromAll:using: or fromAll:using:.

Could you look at this as well?

Cheers,

Doru

On Thu, Oct 30, 2014 at 9:16 PM, milton mamani <[hidden email]> wrote:

Hi Tudor

I update the TreeMapBuilder in order to explore nodes of one tree similar to Sunburst.

The next two images show the execution of this code:

MessageTally on: [ 

b := RTTreeMapBuilder new.

b

extent: 400@400;

weight: [:n | n].

b explore: 1

using: [:n | 

n >= 10000

ifTrue: [ { } ]

ifFalse: [ (n*10 to: (n+1) * 10 - 1) ]].

b open ] 

<image.png>

<image.png>

Cheers,

Milton

2014-10-30 7:25 GMT-03:00 Tudor Girba <[hidden email]>:

Hi Milton,

I cannot see the pictures that you sent, but I did try your algorithm and it does not solve yet the original problem.

Try this:

MessageTally on: [ 

b := RTTreeMapBuilder new.

b

extent: 400@400;

weight: [:n | n].

b explore: 1

using: [:n | 

n >= 10000

ifTrue: [ { } ]

ifFalse: [ (n*10 to: (n+1) * 10 - 1) ]].

b open ] 

On my machine it takes some 10s, and half of it is spent on buildEdges (see the attached picture).

<tally-explore-using.png>

This part we can dramatically optimize by simply creating the nodes during the createNodes:using: instead of creating edges only to afterwards look them up. For this, you would need to extend the recursive traversal by also passing on the parent node so that you can create directly children of it. 

The Sunburst version does that. For example, this:

MessageTally on: [ 

b := RTSunburstBuilder new.

b explore: 1

using: [:n | 

n >= 10000

ifTrue: [ { } ]

ifFalse: [ (n*10 to: (n+1) * 10 - 1) ]].

b open ] 

shows that the traversal is much faster (some 150ms).

Could you try to implement like that?

Cheers,

Doru

On Wed, Oct 29, 2014 at 10:46 PM, milton mamani <[hidden email]> wrote:

I update the code of RTTreeMapBulder to get this:

| b |

b := RTTreeMapBuilder new.

b

extent: 400@400;

nodes: (RTShape withAllSubclasses);

edgesFrom: #superclass toAll: #yourself;

weight: [:el | el numberOfMethods ].

b shape 

if: [ :cls| Smalltalk includesKey: (cls name,'Test') asSymbol ] fillColor: Color lightGreen;

if: [ :cls | cls name endsWith: 'Line' ] fillColor: Color purple;

if: [ :cls | cls hasAbstractMethods ] fillColor: Color red.

b popup text: [ :cls | cls name, ' number of methods: ', cls numberOfMethods asString ].

b open.

This is now in Roassal repository.

You can also build something like this:

        | b |

b := RTTreeMapBuilder new.

b

extent: 400@400;

weight: [:m | m tally ].

b explore: (MessageTally new

spyEvery: 1

on: [Nautilus open]; yourself)

using: #children.

b open.

Cheers, 

Milton

2014-10-29 15:49 GMT-03:00 Tudor Girba <[hidden email]>:

I believe the idea of the API is to simulate the Mondrian API, but it does not make much sense in the context of the treemap because we only need to build a tree, so we do not need to provide a mechanism with which to build a graph.

Actually more than half of the time is being spent on precisely constructing the tree. We should simply have a construct that acts like we have in the DeepTraverser:

http://www.humane-assessment.com/blog/traversal-enabled-pharo-objects/

Cheers,

Doru

On Wed, Oct 29, 2014 at 7:42 PM, Alexandre Bergel <[hidden email]> wrote:

I do not understand how it works. 

I tried the following:

| b |

b := RTTreeMapBuilder new.

b

extent: 400@400;

nodes: (RTShape withAllSubclasses);

edges: #superclass;

weight: [:el | el numberOfMethods ].

b build

Apparently edges is not well specified. What I should provide in edges?

Alexandre

-- 
_,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
Alexandre Bergel  http://www.bergel.eu
^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.

On Oct 29, 2014, at 12:44 PM, milton mamani <[hidden email]> wrote:

The code is now in Roassal2 repository

2014-10-29 4:27 GMT-03:00 Tudor Girba <[hidden email]>:

Where is the code?

Doru

On Wed, Oct 29, 2014 at 2:55 AM, Tudor Girba <[hidden email]> wrote:

Great. I will look at it tomorrow. I am also interested in the time complexity.

Doru

--
www.tudorgirba.com

"Every thing has its own flow"

> On 28 Oct 2014, at 21:40, Alexandre Bergel <[hidden email]> wrote:
>
> Can someone have a look at this?
>
> Cheers,
> Alexandre
>
>
>> On Oct 28, 2014, at 5:14 PM, milton mamani <[hidden email]> wrote:
>>
>> Hi,
>>
>> I been working in RTTreeMapBuilder,
>>
>> Try to run the next piece of code
>>
>> .=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=
>> | b |
>> b := RTTreeMapBuilder new.
>> b
>>    extent: 400@400;
>>    nodes: (1 to: 76);
>>    edges: ((1 to: 76) collect: [:n | (n //10) -> n]);
>>    weight: [:n | n].
>> b shape
>>    if: [:n | n even ] fillColor: Color lightGray;
>>    if: [ :n | n odd ] fillColor: Color purple.
>>
>> b open.
>> .=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=.=
>>
>> to get this preatty visualization.
>>
>> <image.png>
>>
>> Let me know what you think about it
>>
>> Cheers,
>> Milton
>>
>> 2014-10-28 14:41 GMT-03:00 Tudor Girba <[hidden email]>:
>> Great. This is the only missing thing :)
>>
>> Cheers,
>> Doru
>>
>>
>> On Tue, Oct 28, 2014 at 1:28 PM, Alexandre Bergel <[hidden email]> wrote:
>> Just to let you know that we are working with Milton on a new treemap.
>>
>> Cheers,
>> Alexandre
>>
>>
>>> On Oct 22, 2014, at 1:14 AM, Tudor Girba <[hidden email]> wrote:
>>>
>>> We "just" need a treemap :)
>>>
>>> Doru
>>>
>>> On Wed, Oct 22, 2014 at 2:24 AM, Alexandre Bergel <[hidden email]> wrote:
>>> Ok, we will pay attention to this
>>>
>>> Alexandre
>>>
>>>
>>>> On Oct 21, 2014, at 5:24 PM, Tudor Girba <[hidden email]> wrote:
>>>>
>>>> http://www.humane-assessment.com/blog/system-nesting-map
>>>>
>>>> Doru
>>>>
>>>> On Tue, Oct 21, 2014 at 10:17 PM, Alexandre Bergel <[hidden email]> wrote:
>>>> What is SystemNesting ?
>>>>
>>>> Alexandre
>>>> --
>>>> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
>>>> Alexandre Bergel  http://www.bergel.eu
>>>> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.
>>>>
>>>>
>>>>
>>>>> On Oct 19, 2014, at 4:24 PM, Tudor Girba <[hidden email]> wrote:
>>>>>
>>>>> Hi,
>>>>>
>>>>> There are only a couple of things left to do in the migration:
>>>>> - Fix the UML diagram to properly show attributes (we need another transformation to show the name of an attribute)
>>>>> - Reimplement a treemap visualization in Roassal2 and have the SystemNesting ported to it
>>>>> - Reimplement the GLMBrowser>>viewTree visualization
>>>>>
>>>>> Anyone interested in helping?
>>>>>
>>>>> Just a reminder: these are pretty much the main things that are still left to do in order for us to release Moose 5. Once we release it, we can move to Pharo 4 and benefit from a larger exposure and feedback for all the tools built around Moose.
>>>>>
>>>>> Cheers,
>>>>> Doru
>>>>>
>>>>>
>>>>> --
>>>>> www.tudorgirba.com
>>>>>
>>>>> "Every thing has its own flow"
>>>>> _______________________________________________
>>>>> Moose-dev mailing list
>>>>> [hidden email]
>>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev
>>>>
>>>>
>>>> _______________________________________________
>>>> Moose-dev mailing list
>>>> [hidden email]
>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev
>>>>
>>>>
>>>>
>>>>
>>>> --
>>>> www.tudorgirba.com
>>>>
>>>> "Every thing has its own flow"
>>>> _______________________________________________
>>>> Moose-dev mailing list
>>>> [hidden email]
>>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev
>>>
>>> --
>>> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
>>> Alexandre Bergel  http://www.bergel.eu
>>> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.
>>>
>>>
>>>
>>>
>>> _______________________________________________
>>> Moose-dev mailing list
>>> [hidden email]
>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev
>>>
>>>
>>>
>>> --
>>> www.tudorgirba.com
>>>
>>> "Every thing has its own flow"
>>> _______________________________________________
>>> Moose-dev mailing list
>>> [hidden email]
>>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev
>>
>> --
>> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
>> Alexandre Bergel  http://www.bergel.eu
>> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.
>>
>>
>>
>>
>> _______________________________________________
>> Moose-dev mailing list
>> [hidden email]
>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev
>>
>>
>>
>> --
>> www.tudorgirba.com
>>
>> "Every thing has its own flow"
>>
>> _______________________________________________
>> Moose-dev mailing list
>> [hidden email]
>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev
>>
>>
>> _______________________________________________
>> Moose-dev mailing list
>> [hidden email]
>> https://www.iam.unibe.ch/mailman/listinfo/moose-dev
>
> --
> _,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:
> Alexandre Bergel  http://www.bergel.eu
> ^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;._,.;:~^~:;.
>
>
>
>
> _______________________________________________
> Moose-dev mailing list
> [hidden email]
> https://www.iam.unibe.ch/mailman/listinfo/moose-dev

--

www.tudorgirba.com

"Every thing has its own flow"

_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.iam.unibe.ch/mailman/listinfo/moose-dev

_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.iam.unibe.ch/mailman/listinfo/moose-dev

_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.iam.unibe.ch/mailman/listinfo/moose-dev

--

www.tudorgirba.com

"Every thing has its own flow"

_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.iam.unibe.ch/mailman/listinfo/moose-dev

_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.iam.unibe.ch/mailman/listinfo/moose-dev

--

www.tudorgirba.com

"Every thing has its own flow"

_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.iam.unibe.ch/mailman/listinfo/moose-dev

_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.iam.unibe.ch/mailman/listinfo/moose-dev

--

www.tudorgirba.com

"Every thing has its own flow"

_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.iam.unibe.ch/mailman/listinfo/moose-dev

_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.iam.unibe.ch/mailman/listinfo/moose-dev

--

www.tudorgirba.com

"Every thing has its own flow"

_______________________________________________
Moose-dev mailing list
[hidden email]
https://www.iam.unibe.ch/mailman/listinfo/moose-dev