Tag: opensource

Google OS Open Source

it has long been argued that the google os, particularly MapReduce and GFS, is google’s real competitive strength. yahoo, meanwhile, is paying developers to develop clones of these. with seeming consolidation on a common computing platform, and ever-rising data center expenses, you gotta wonder how much sense it makes for the big 3 to duplicate all that CAPEX. they might be better off outsourcing their datacenters, and sharing some base datasets, such as a crawler cache (kinda like the feedmesh network).

the outsourced company, on the other hand, would end up running a grid with several million nodes and could optimize running costs overall, by using very low power servers, running on an opensourced processor architecture.

content wire interview

i recently did a new interview with paolo di maio of content wire.

Gregor J Rothfuss has been observing and working with content management tools for many years. We catch up with him over the internet to talk about open source, a subject increasingly recurring

How do you see the business model for OS developing?
In services, more and more. there are some excellent open source venture blogs that track this question in great detail: Open Source Strategies and Asay

Os tools are not very usable. Why is that? Is it just still too early in their evolution?
This held true, traditionally, but has recently been true less and less. One reason is that software per se is no longer interesting, and increasingly, developers are aware of usability issues, and a nice UI can now be considered quite sexy. This is mostly due to some well-publicized web applications that utilize AJAX technologies. At first, OSS was often in catch-up mode and had to quickly fill in holes in functionality. now that functional parity to commercial products is being reached, the focus has often shifted to these more subtle qualities.

Plus, more and more hackers understand the value of design, as exemplified by Apple who are taking geek toys and slapping nice UI on top, or Firefox, which is essentially a slimmed down Mozilla.

Only in its Firefox incarnation has full success come.

Post-sourceforge world

People have asked the question “What if SourceForge disappeared?? for years now, but I have to wonder if we should be asking this question again. Now, SourceForge has its warts, but it’s ultimately a beneficial service. And, even if they did disappear, it’s highly unlikely that the open source movement would be handicapped for any real length of time. But here’s why I ask the question:

phil goes into some more detail, wondering whether GOOG or YHOO might be prepared to take over. Maybe the woes of sourceforge can bring some long-needed fresh air though:

I think that the problem with SourceForge is that they are providing 1999-era functionality based on a business model that really is not much more than an afterthought after the collapse of their hardware business. Consequently, the core functionality in the SourceForge project hasn’t changed all that much in the past 6 years. All the projects on SourceForge are effectively partitioned… we don’t see any tools for figuring out code reuse possibilities or anything particularly innovative.

3 years ago, i researched the state of the art of open source production, and developed a matrix to map activities, actors and tools. To say that there are many areas of improvement in the way open source software is produced is an understatement. The obvious observation that there are a power laws in effect with regards to quality and popularity of a project makes me wonder what can be done to improve life for the countless small projects out there that have neither their act together code-wise, nor any audience. A recent study found that 81% of sourceforge projects are inactive, and only about 0.05% innovative.

A considerable subset of these projects deserves to do better on both fronts with the right tooling. Done correctly, a post-sourceforge integrated site could act as a large-scale lab for novel collaboration and software engineering techniques. Tool vendors might be willing to integrate their technology in return for widespread usage and name recognition, and the rest of us might finally break free of the anachronisms of mailing lists and other 1980s-era solutions.

Furthermore, the site could be made to emit statistical data for open source research. Most academic papers in the field already look at sourceforge anyway, if they had a way to get better data, they might actually arrive at some useful conclusions, including reuse patterns, social network analysis, and many more. Such a site could therefore be a down payment on discovering the finer points of peer production, slated to become ever more important in the larger economy.

Open-source business infrastructure

commerce.net has set up a lab to explore decentralized electronic markets.

We propose developing an open-source toolkit for creating markets, called Zocalo, in order to catalyze broader adoption of markets in academia, industry, and throughout society. We are primarily interested in prediction markets, which allow traders to buy and sell securities that pay out based on the outcome of some future event, but Zocalo would also be useful for creating markets in other goods.

6 reasons from the linked paper why prediction markets are a good choice for decision support applications:

  1. They give continuously updated dynamic forecasts
  2. They aggregate information across traders
  3. These markets give unbiased, relatively accurate forecasts well in advance of outcomes
  4. They can outperform existing alternatives
  5. market dynamics overcome biases that individual traders may have, effectively eliminating [the biases] from forecasts
  6. They can be designed to forecast a variety of issues and provide a variety of types of information

with luck, efforts like this (or ubl) might make hacking on business problems sexy / interesting enough to foster new open source communities. ideally, all these efforts nudge corporate entities towards peer production (bye bye HR raffles, bye bye conference room campers)

RAIDb

just learned about RAIDb

RAIDb stands for Redundant Array of Inexpensive Databases. This acronym has been used in reference to the RAID (Redundant Array of Inexpensive Disks) concept that achieves scalability and high availability of disk subsystems at a low cost. RAIDb aims at providing better performance and fault tolerance than a single database by combining multiple inexpensive database instances into an array of databases.

C-JDBC is a database cluster middleware that implements RAIDb, LGPL licensed.

State of X Windows

Jim Gettys, one of the main developers of X Windows, gave a talk about the future of X11 at MIT. The present had some nasty surprises in store, when it took Jim a considerable while to make X11 talk to the projector 😉 It is gratifying to know that even the gurus sometimes fail with the “simplest” tasks.
Rough notes from his talk:
X’s flaws
X’s font architecture fundamentally flawed
Inadequate 2D graphics
Accessibility difficult to implement
No Eye candy
Full integration into 3D environment
Collaborative shared use of X
Cairo brings extremely high quality graphics to free software. Client side fonts are the biggest change in X since 1988.
Try turning on hinting in Freetype (patent problem prevent it from being on by default.)
Lessons learned:
X is but one component of a complete desktop environment
no more flashing: double buffering everywhere
To preserve traditional X flexibility, new Compositioning Manager is introduced. Solved via indirection, copied to visible frame buffer by compositing manager. Apps never draw the screen, compositioning manager applies whatever effects are appropriate and does the screen drawing.
Mouse input needs to be captured and transformed for accessibility. XEvie by Sun does this.
4 new extensions: XFixes, Damage, Composite, XEvie
Eye candy is designed in a way that institutes an upper bound for computation cost. This is a very nice design. (Bound is human perception)
Demo with drop shadows, fade in / out. “A 1000 eye candy flowers will bloom, most will stink, but some will smell nice. Expect a cambrian explosion of eye candy.”
X becomes just another GL application. Croquet project highly recommended.
Jim evoked visions of millions of people sharing 3D spaces with it.
Efforts underway to make X build more easy. Switch to autotools underway. 90% of server round trips can be eliminated. Network compression for synthetic images can do 300x (needed for competition with RDP, Citrix) Session migration needs Xlib support for connection loss. There is a migration prototype done in GTK+ 2.2 One goal is to make input devices network transparent (on the way to X11 entertainment center 🙂
“Network audio solutions are all very lame”
“XFree86 is dead”, most developers moved to X.org. It moved from industry consortium to open membership. X11R6.8 includes X fixes, Evie, Damage, Composite, shipping now. New releases happening much more frequently now.