Viewing posts for the category presentations
From the outset, dataflower is meant to be a practice based research project. My main motivation for it came from the experience of many years developing digital art installations with diverse teams, noticing many recurring technical and soacial patterns and thinking "there has to be a better way to approach these problems". Most of my research on this project so far has involved creative practice, but mostly in a non-participatory way, through observation, interviews and questionnaires. To actually get involved in creative projects and use practice as a way to shape the development of dataflower, I first need to have a minimal working version which actually does something interesting and useful.
I now have such a prototype of the whole system which I can use as a foundation for more active practice-based research, but I still need people to experiment and to try building projects with it and give me feedback, so I need to start meeting and working together with them more regularly, so i came up with the idea of organizing data gardens.
A data garden is an open collaborative development event for all those interested in learning, working with and improving the dataflower tools and on reflecting about the issues related with parallel, heterogeneous and distributed computing in digital art practice. The focus of each event will depend on the participants interests, but my plan is to have a specific topic to discuss or a new feature to test for each session, which should happen at least once a month.
Last saturday, July 19, I got together with a group of digital artists and researchers at Audiência Zero in Matosinhos, Portugal, to perform a pilot test for these sessions. Since this was a first and the tools are still a bit rough around the edges, I started with a small presentation to present the project and its motivation and to elicit feedback from the participants. I then presented a brief tutorial of FBP concepts using microflow and gave a brief overview of the upcoming libflow library. Next I presented a prototype of the dataflower environment and we had a brainstorming session on mental models for complex systems and collaboration tools.
Many thanks to Audiência Zero for hosting the meeting and to all who participated for their interest and wonderful feedback. The next data garden session is scheduled to take place in Coruña, Spain, from 17 to 21 of September as a work node part of the Artropocode residency, and you are most welcome to join us.
If you want to host or participate in a future data garden event, get in touch.
I was lucky enough to attend Codebits again this year, and the event was awesome as usual. In the middle of all the things to see and do at the event I was able to present a talk again, this time focused on the topic of Flow-Based Programming. For those whose could not come to the event, here is the original talk description:
As software developers, every day we are faced with ever more complex systems to run our applications. Single machines are not enough. Now we need to orchestrate hordes of multicore CPUs, GPUs and DSPs to run our applications in a scalable, distributed way and this is hard.
Flow-based Programming is a programming methodology created by J Paul Morrisson at IBM in the 70s in order to solve everyday challenges he faced when developing large scale business applications. Based on a solid theoretical foundation and battle-tested on real-world applications (with one of them running continuously for the last 40 years), FBP is making a comeback as an effective model to reason about and implement data intensive, distributed applications, as can be seen by the growing interest in tools like NoFlo and Storm.
In this talk I will present the history and fundamental concepts of Flow-based Programming, and how it is different from other models like dataflow, functional and object-oriented programming. We will then use noflo to develop a small heterogeneous application using this methodology.
And here is the talk video:
I think the talk went well and that I managed to get the main ideas across, although looking back it probably sounded a bit confusing because I couldn't see my notes during most of the talk. I'll see if I can expand those notes a bit and post them as a follow-up to this post. In the mean time do check the other Codebits 2014 talk videos, there's lots of good stuff there.
Here's hoping to be able to present again next year to showcase dataflower to the Codebits crowd.
Every year, the UT Austin|Portugal Colab program organizes the Future Places festival in Porto, Portugal. The first day of the festival usually starts with a research symposium where the program's PhD students (like myself) present the state of their work and gather feedback from their colleagues and the program's faculty. It is always a great opportunity to meet new and old friends and to learn about everyone else's research experience.
I presented the work done so far in dataflower, a condensed version of my research proposal, and some preliminary ersults of the need finding study I'm currently working on. I received some very good feedback and I'm definitely looking forward to participate again next year.
In order to fulfil my PhD requirements, last January I had to present a research proposal outlining my thesis topic, a rough state of the art and a research plan. Last week I successfully presented my proposal to my thesis committee and had the good fortune of having Professor Brad Myers, head of Carnegie Mellon University's Natural Programming Project, as the main jury member.
I really enjoyed the discussion that followed the presentation and I feel that it helped me focus on what could be the core contribution of my work, namely to better understand the role collaboration plays in modern digital creativity and to focus on designing tools that better support this creativity.
I would like to thank Professor Myers, all the committee members and everyone who has discussed this work with me so far, your support has been precious. Now it's time to make good on the promise of that research plan and get to work.
LGM 2013 was held in Madrid at the brand new Medialab Prado building, a fantastic venue which besides a large auditorium provided lots of work space for the developers present. The program was divided into workshops and work sessions in the morning, conferences during the afternoon and social events in the evening. Many of the participants there were also going to stay for the following week for the Interactivos production workshop.
With so many activities and so many interesting people there, it was not possible to attend every session, but here are my notes on some that caught my attention:
Camille Bissuel presented Mikado, a graph-based image editor which has a structure similar to the one I'm proposing for dataflower. It is based on a C++ dataflow library called Tempi and has a web-based UI. I had a chance to briefly talk to the presenter and there are many good possibilities for future collaboration. Tempi looks like a very mature and well designed library and when the time comes I'd like to contribute to it and support it as a backend in dataflower.
Tom Lechner presented his Laidout tool which he's been developing for some time now to do the layout of his own publications. The tool implements many really interesting ideas, eschewing most of the traditional idea of tools in favor of direct manipulation of objects in the canvas. On another related talk, Tom also presented some thought-provoking ideas on how we could share tools between Free Software graphics applications.
Dave Crossland and Ben Martin talked about the current state of Fontforge and demoed an impressive real-time collaboration feature that they would be working on during the conference and the Interactivos workshop.
Ale Rimoldi, inspired by how UNIX shell pipelines work, presented his thoughts on how we should be able to connect graphics applications to build complex workflows, similar to what we can already do with audio applications via JACK.
Danji Vasiliev presented his Hotglue project, a web framework that allows users to direcly edit their web pages in the browser through direct manipulation. He also presented his plans to create Superglue, a distributed platform web publishing platform based on Hotglue.
You can find all the talk videos in the Medialab-Prado multimedia archive.
As for my talk, I'm pretty happy with how it turned out and got really good feedback from the people attending, but I'll let you be the judge of that. Here is the video and the slides of my presentation:
LGM 2013 was an incredible experience for me, as I've had the chance to meet some good friends and some of the people that develop many of the tools I use and love. I also got the chance to take the pulse on what the Free Software community is doing to support creativity, and regarding the future of these tools I can tell you, the future is Free (as in Freedom).
I just received word that my talk proposal for Libre Graphics Meeting 2013 has been accepted. The selected theme for this year is "Future Tools", so this will be a fantastic time to present my work, get in touch with the Libre Graphics community and discuss the future of Free (as in Freedom) creative programming tools.
Hope to see you in Madrid next April. Meanwhile you can already get involved in the conference and join the ongoing discussion that is being used to generate the posters for the event.
Update: the conference program is up.
SAPO Codebits is the largest technology event in Portugal. Every year, around 800 developers and technology enthusiasts apply for a chance to geek out, present their pet technologies and compete in a furious 48h free-for-all hackathon with awesome prizes.
I've been to Codebits many times but never felt I had anything inteersting to propose as a talk. In 2012, realising how little attention was being paid to parallelism in the conference and how little most developers knew about the subject, I gathered my notes and proposed an overview talk about the rise of heterogeneous parallelism, parallel algorithm design and OpenCL.
Here is the video: