Well, how can I add a model... there are now menus and the list is also not editable…?
Currently, CSV tables are not hooked into Moose. I have no idea if someone has worked on this or not. On my side, I have never hooked CSV table to Moose because I have never seen a need for this.
Consider the following script:
-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-= |tab ds b | tab := RTTabTable new input: 'http://earthquake.usgs.gov/earthquakes/feed/v0.1/summary/2.5_month.csv' asUrl retrieveContents usingDelimiter: $,. tab removeFirstRow. tab replaceEmptyValuesWith: '0' inColumns: #(4 5). tab convertColumnsAsFloat: #(4 5).
b := RTGrapher new. ds := RTDataSet new. ds interaction popup. ds points: tab values. ds dotShape color: (Color blue alpha: 0.3). ds x: [ :row | row at: 4 ]. ds y: [ :row | row at: 5 ]. b add: ds.
b maxY: 8. b maxX: 700.
b axisX title: 'depth'. b axisY title: 'magnitude'. b build. -=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=-=
It visualizes seisms in plain Roassal, without the need to have it hooked into moose. This is a singular example. If I would deal with Seisms the all day long, then yes, I would surely hook it into moose (e.g., extending the Moose meta model, adding an importer in the Moose Panel, …).
At that stage, I have never seen the need.
Regarding the data, I just have a 300MB csv file that I imported into a data structure in Pharo. This blew up the image to 780MB.
Do you really need to keep all the cells at the same time? If this is the case, when you will probably need to work on the way to approach this case study. Several ways are possible: - You can use fuel to load and unload part of the CSV table on the disk. This should not be difficult at all. - You can keep in the image only the index of the CSV row, and access to it when you wish to get more data from it.
I am sure there are other way to deal with such a large file. It all depends on what the file contains.
Doing calculations and visualization in this image is not doable. I'll need to span that off to a different image or do stuff step by step.
Again, this is something that is not difficult to do. Which kind of calculation? You can either use OpenCL or span other images. Again, we have solution for this. For example, feel free to get in touch with Juan Pablo Sandoval, my PhD student. He has worked quite a lot on spanning new images.
I was just wondering what you guys do since you might have this problem on a daily base. I can already tell, that the image become very slow once a visualization gets too big. Does Roassal use double buffering for the moving and scaling of the visualization?
Roassal is indeed becoming slow when you reach around ~ 400 000 elements on screen. This is not really a surprise because there is no optimization. However, all is made to hook strategy for large amount of data.
Do you know if the team in Argentina analyses the network from within Pharo?
Yes, all within Pharo.
Alexandre
Have a great weekend! Sebastian
Hi Sebastian!
I am doing my first steps in learning Moose and also Roassal. Is it possible to start Roassal Easel and Moose Panel from the world menu and add a model afterwards?
Yes
How would one visualize data with the help of Roassal which underlying data is too big for the Pharo image?
Well… it depends. It all depends on the granularity of your visualization.
Is there an example available how one could stream information on a Roassal view?
There is not such a thing so far. Which kind of data do you have? It all depends on the data, their structures. If you are simply interesting in visualizing some metrics, then this is easy. If you are interested in visualizing structure of a large amount of objects, then we need to be careful.
Did somebody already visualize a LAN network? What would be a good framework/project to do such analysis?
People in Argentina are using Roassal to monitor network packets for a TV digital project.
Cheers, Alexandre _______________________________________________ Moose-dev mailing list Moose-dev@iam.unibe.ch https://www.iam.unibe.ch/mailman/listinfo/moose-dev
Moose-dev mailing list Moose-dev@iam.unibe.ch https://www.iam.unibe.ch/mailman/listinfo/moose-dev