Monthly Archives: March 2014
Just watched the keynote at Google Cloud Platform Live event, which was held several days ago. It nicely includes many demonstrations along with the main presentations. AWS is leading this field at the moment, but google seems re-attacking with drastic discounts and many new features.
At around 40:00, it demonstrates the online Developers Console features to easily deploy codes and debug errors at production environment.
At around 52:00, it demonstrates new Managed Virtual Machine features. It allows to customize the virtual machine configuration using XML configuration file, like listing-up apt packages to be installed. It sounds simpler than managing chef cookbooks (which I’ve been struggling these days). Though it might cause some restrictions installing the latest or minor packages, it may be a nice choice in the case your server requirement fits well.
Anyway, it’s good to have many options for users, and the PAAS/SAAS fields are yet getting more and more interesting.
The aboves are introductory video descriptions for Neutron, which is the OpenStack’s networking component. It’s demonstrated by VMWare engineers, and it’s interesting that VMWare is aggressively involving into the OpenStack community, though they have their own strong platforms. After the basic hypervisors became commodity, virtualize-focused companies seems pursuing different ways to gain the presence in the community.
According to the video, OpenStack’s computing engine (Nova) has basic flat-networking model based on VLAN on physical switches, but it has some limitations in the scalability. Then, Neutron extends the capability to software-based isolated virtual networks, and trying to resolve these issues (it provides agent and plugins to interface with OpenVSwitch).
VMWare is offering proprietary virtual networking like VMware NSX, but it’s exposing integration options with OpenStack, rather than facing into their product platforms. This kind of unification of interfaces would be nice directions for users.
A great interview of Masayoshi Son, at Charlie Rose show. Masayoshi Son is one of the greatest entrepreneur in Japan. This interview reveals some of his strong passion to the information revolution, and the long-term vision behind.
During the Softbank’s growth in Japan, it was sometimes criticized by their aggressive approach. But, Japan would not have gained the broadband network which we now have. It’s interesting to see he’s still extending further to the United States wireless market, and I’m curious how it turns out.
Nice talk about the difference in types at different languages from the scala author.
The following category was used for describing static type systems.
- Type it to the max (Scala, Haskell, OCaml, F#)
- Cutting corners (Dart, Eiffel, Typescript)
- My way or the highway (Go, Pascal, Java4)
Interestingly, each of the category has actively developed ones, like Dart, Go, Scala, along with majority languages like C and Java. It would be indicating the difficulty in optimizing the policy for every problems.
As indicated in the presentation, major part of the scala’s complexity may be coming from the types. When I tried some, understanding type error was difficult. The major pain was to identify how to solve them (how to convert one type to another is sometimes difficult).
dotty seems one trial for alleviating the complexity, and code is uploaded in the following GitHub repository.
Dotty is a platform to try out new language concepts and compiler technologies for Scala. The focus is mainly on simplification. We remove extraneous syntax (e.g. no XML literals), and try to boil down Scala’s types into a smaller set of more fundamental constructors. The theory behind these constructors is researched in DOT, a calculus for dependent object types.
Just created a simple twitter client library as part of learning OAuth. erlang-oauth works just nicely with Twitter REST API.
Also, for http mocking test, added support :httpc module in exvcr, as erlang-oauth uses :httpc. Limited methods are available yet, but I may be adding more later.
An interesing presentation about creativity in machines. It discusses about the meaning of creativity, using musical demonstration with overtone.
Now, computer graphics are everywhere. Most of the movies use them, and sometimes it’s very difficult to distinguish them with the “real” one. Industrial robots are everywhere too, and most of the factories use them, without human instructions or interventions.
Still, the artificial intelligence with creativity seems like not ready yet. Sometime ago, there were some buzz around the robots which interact with humans, like aibo and ashimo. But, there’s no major news around it recently. I grew up with Doraemon; story about a robot came from 22th centry, and I feel a little sad that it’s not coming yet.
In the presentation, various methods to generate “music” are demonstrated. It ranges from basic math functions through the waves sampled from brains. This brings to think what’s good/bad music, but it’s difficult to tell. Sampling from actual musicians and combining them with statistical methods are other approaches, but it can be creative or can be just a cheating. It’s an interesting topic.
As also noted in the presentation, machines don’t understand beautiful and clean codes, and also machines have no idea about the variable names, yet. It’s an interesting viewpoint too.
Is there any future that machines hate accepting minified/uglified codes?