A Report on SCaLE5x
[ Kapil sent in a series of e-mail messages to TAG during SCaLE. I'm enclosing those as well as some followup e-mail in the Mailbag format for the enjoyment of the rest of our readership. - Kat ]
Scale5x Report 1
Kapil Hari Paranjape [kapil at imsc.res.in]
Sun, 11 Feb 2007 09:13:27 -0800
[[[ Hello, I've been attending Scale5x. Here is my first report. - Kapil ]]]
Don Marti spoke about "Going beyond User-Friendly to User-Intimate with desktop tweaking power".
He began the talk with a description of how various platforms and GUIs have evolved the Human Interface Guidelines and this was a bit worrisome since I didn't want to hear much more about those. But I need not have worried---he was just setting up an obstacle so that he could show us the ways around it!
Thomas Adam would have found this talk interesting since the subtext of this first 3/4 of talk was "How can we make Gnome/Metacity behave more like fvwm/scwm/sawfish when we want it to?" Marti gave a number of tips on how to use the tools "xev", "xwininfo", "xlsclients", "xsel" and friends to make it possible to provide a command-line/keyboard/scriptable interface to applications have only dialog boxes and mouse clicks.
In the last quarter of his talk Marti talked about some new ways of interacting with the computer. The simplest of these are the extra keys on laptops which are not handled by the keyboard driver. More unconventional are things like the hard-disk acceleration detection driver on some laptops; you can use this to detect the orientation of your laptop to flip the screen. A rather cool idea was using your hand to cover the wireless antenna and have a program that detects Wi-Fi signal strength use this in some way---for example, as an application launcher or even a "theremin".
Marti then outlined some dangers of this way of tweaking the interface:
(a) Your computer truly becomes personal. Soon no one else can use it and you can't use someone else's computer either! (b) This kind of programming really sucks up time without necessarily contributing to the community ... unless you give a talk about it or write an article explaining things.
All in all a fun talk and a fun way to start the day.
[ Thread continues here (3 messages/4.56kB) ]
Scale5x Report 2
Kapil Hari Paranjape [kapil at imsc.res.in]
Sun, 11 Feb 2007 09:15:28 -0800
Roger Klorese from Xen spoke about "Xen: The Open Source industry standard for virtualisation".
For those of our readers who don't know this yet, "Xen" is a system for running multiple "virtual" systems on one box. As the commonly available systems today have more and more unused resources, running virtual systems is one way to get more out of your box.
The mechanism used by "Xen" is called "paravirtualisation" which I will try to explain briefly below to the best of my understanding.
There is always a "master" system which we can call "host"; the others are called "guest" systems. In older virtualisation techniques, system calls generated in the "guest" system were trapped and passed to the "host" to be handled. This trapping was done by generating a fault whenever such a system call was executed.
In paravirtualisation this fault generation is avoided in one of two ways; in either case this leads to smoother running "guest" systems. One is by re-building the kernel of the "guest" virtual system so that it interacts with a certain well-defined "API" for system calls called the "Hypervisor API". The other is through "hardware assist" where the CPU is given a "Hypervisor Table" by which it re-directs calls to the "Hypervisor". In either case there is a Hypervisor layer which is not a full system but merely a kind of system-call router. Thus the Hypervisor could be made to route the system call to the "host" or could also be made to route the system call directly to the hardware---the hypervisor configuration tools can be used to dynamically change this routing information.
The speaker clearly had a problem about where to pitch his talk---the jargon did not grow steadily but came and went in waves. While some of "Xen" is available as Free and Open Source there is also a product that is released by the the same group. The lines between what are features of the Open Source version and the product version were often not clearly pointed out, which made this occasionally sound like a marketing talk. This feeling was accentuated by digs at the "opposition".
In summary, the whole game of "virtualisation" is clearly an interesting one to get involved in during the coming year(s). The talk could have been better.
Scale5x Report 3
Kapil Hari Paranjape [kapil at imsc.res.in]
Sun, 11 Feb 2007 09:16:40 -0800
Jono Bacon of Canonical spoke about "How to Herd Cats and Influence People". This was basically a talk about how to build the community around an open source project.
Jone clearly had a lot of skill in doing this as he got the audience chanting and interacting with him---he even encouraged some Gentoo hecklers! He explained his experience with Ubuntu and Canonical and creating the community around the Ubuntu project---this is his primary purpose at Canonical. According to him there are four important aspects of this work:
1. Find good people. 2. Organise them in teams. 3. Make an overall structure which keeps these teams working with each other. 4. Create a mechanism to bring in new people.
Though this talk was very entertaining in many ways I can't honestly say that I brought away much more than that from it.
In particular, the aspect of "Herding Cats", which I thought referred to the fact that these are all quite smart people who consequently do not "herd" easily, was not discussed much. Is the Ubuntu community less visibly "flame"-y than the Debian or Gentoo community because it is young or is it something in the way they organise themselves? A discussion of this question is what I would have liked to hear.
Scale5x Report 4
Kapil Hari Paranjape [kapil at imsc.res.in]
Sun, 11 Feb 2007 09:17:36 -0800
Alex Ionescu spoke about "ReactOS--bridging the gap between Windows and Open Source". He is the lead developer in a project to construct a clean-room implementation of an open source replacement of the Windows NT kernel. This replacement is the core of ReactOS.
Alex explained that the motivation for this work was from many directions. There is the perennial inter-operability issue; to help us make use of documents and/or environments that are meant for that 90-95% of the computer user world that is still on Windows. There is the research interest in a operating system that has a fundamentally different design from the monolithic Linux or BSD kernel that we are used to. There is also the desire to play games that are written for Windows-only, or more generally to run that one application that some users have not found a Free and Open Source alternative for. He didn't mention the users who might need ReactOS when MicroSoft stops supporting Windows XP and starts requiring users to upgrade to Vista.
Alex explained that the kernel is working fine but there are still a lot of issues to be sorted out---especially with the layers over the kernel. In particular, the GUI of Windows is driven through the "Win32K" layer for which there is currently no complete replacement in ReactOS. However, they are currently able to run Open Source applications on ReactOS --- including OpenOffice, Evolution and Firefox. Unfortunately, they cannot run MS-Office, Exchange or Explorer just yet.
Alex tried to motivate the audience to get involved with this project, as it has a number of benefits to the Free and Open Source community in addition to the ones mentioned above.
1. Programs get testing and debugging in an environment that is different from the standard one---this points out some flaws in the login. 2. Programs meant for Windows can be debugged and traced which is much more difficult to do with the native environment without signing various NDA's and buying expensive tools.
This was a well organised and well presented talk. I went for it because of my interest in operating systems and also in order to underline my support for the hard work done by a group of dedicated developers. It is difficult to get rid of the feeling that this project will get little more than that from most of us.
Scale5x Report 5
Kapil Hari Paranjape [kapil at imsc.res.in]
Mon, 12 Feb 2007 10:08:41 -0800
Chris Dibona spoke on "Open Source licensing, a review with some corner cases". Chris Dibona is now working for Google so I thought it would be a different point of view from that which Rick Moen has been sharing with us. In short, it was not (though Chris did not use the term "badgeware")--Rick had prepared me too well.
[[[ Kapil is referring to (OSI, GAP, and "Exhibit B" licences) and these followup discussions--(Talkback:134/moen.html, Issue 136 Mailbag, Part 2). - Kat ]]]
For those who have been following Rick's postings to TAG this will be old hat but let me try to review what Chris said anyway. He explained why trying to get credit in various ways:
1. By a direct request for payment.
2. By requiring explicit attribution ("badgeware").
3. By requiring acceptance of philosophy ("don't use this for
the military").
4. By requiring the purchase of specific hardware.
and so on, leads to licenses which are not open source. To those who need any of that Chris' advice was "Don't try to be Open Source---it's OK! Just don't claim to be Open Source either!"
He had two bits of advice. One was to read Professor Ed Felten's deposition to the court regarding the relation between code and free speech. The other was that we should all read the text of the GPLv3.
All in all a great talk and a great way for me to start day 2 --- (Yes I was late getting there).
Scale5x Report 6
Kapil Hari Paranjape [kapil at imsc.res.in]
Mon, 12 Feb 2007 12:47:57 -0800
Ted Haeger and Erin Quill from Novell/SUSE spoke on "Linux, Virtualisation and Virtualisation Management".
I wrote earlier how there were talks on Xen and virtualisation and given some explanation of how I understand these concepts. In this talk the speakers wanted to show us how easy it is to install a new virtual machine under SUSE Enterprise Linux and how easy it is to manage such a machine.
The installation was really easy but I wished they had explained the steps a little more than saying "you point here" and "you click there" and (but for an occasional "cluck" when the error box appears) you are done. Just talking through the steps that are happening under the hood so that we could visualise what was happening would have helped those of us who wanted something a bit more technical.
The second part, about virtual machine management, was also quite impressive with regard to the ease of management via a graphical interface. All the resource allocation of and for virtual machines was done from a point-click-drag-drop kind of interface. It is clear that SUSE have fully absorbed the GUI paradigm for system administrators that Novell perhaps pioneered.
It is perhaps also clear that I do not particularly like this paradigm. If you do not know what is happening under the hood and cannot tinker with it a bit, you should probably not be doing system administration---IMHO. Still, the talk showcased some impressive prettying-up of the Xen virtualisation system.
Scale5x Report 7
Kapil Hari Paranjape [kapil at imsc.res.in]
Mon, 12 Feb 2007 10:09:02 -0800
The last talk (for me) at Scale5x was the one by Don Armstrong on "From Sarge to Etch". Don Armstrong is a Debian Developer who among other things is one of the people managing the Bug Tracking System (BTS).
Now any review by me of a talk by a Debian person is probably very likely to be biased, but I thought it was a very rich talk. If anything, Don had too much material (which is a common error of graduate students!). So let me summarise the main points:
1. How does Debian release? What users can do to help. Some tools to help users help Debian. 2. Using "aptitude". Why "aptitude" surpasses "apt-get". 3. Upgrading from Sarge to Etch. 4. Pointers to obtaining information about Debian, Debian developers and Debian packages.
One important point he made which may seem contradictory is that Debian users who keep wanting Debian to release "etch" as stable will actually benefit from checking it on a "test" machine and filing bugs. While this may increase the RC bug counti, it will improve the quality of the release. Furthermore, fresh information on a bug can often help track it down. So his message to all Debian users was-- keep finding and filing those bugs. (Hint: Use "reportbug" when you do.)
The heart of "apt-get" is "libapt" and the former was meant as a "demo" program--which achieved "world domination". The newer curses and command-line program "aptitude" is a meant as a more complete interface to "libapt". In particular, it can do everything that "apt-get" can do and more. One important addition is that it can give hints for conflict resolution. Another is that it can keep track of packages that are installed only as dependencies--in other words it incorporates some features of "deborphan"/"debfoster". Then there are pattern searches. All this with good documentation. Check out "aptitude" and you will enjoy it--the "etch" version is even better!
The entire text of Don's talk and host of other information (like the re-usable bits of his "rc" files) can be found on his web site at http://www.donarmstrong.com. (I was surprised and pleased to see that his "graphical" interface is "ion" and his chosen method of creating a presentation is "latex-beamer".)
A great way to end.
Scale5x Report summary
Kapil Hari Paranjape [kapil at imsc.res.in]
Mon, 12 Feb 2007 10:09:16 -0800
One of the points of going to such conventions is to meet people.
I met faces that were only familiar from photographs--our editor and our TAG editor. Indeed, I met Ben and Kat on Day 2. That was really great but we could not chat much given the tempo of talks and so on.
Part of the adventure of getting to Scale5x was exactly that--getting there. I had decided that since it was only myself that was being transported--no other people and no equipment--it was time to be environment-friendly and use Public Transportation. "Public Transportation in LA? You must be kidding!" At least that was the response of many people I talked to. In the end, the journey by bus and metro was rather easy. It was long but far from boring. I had some mathematics to keep me company, but there was all the different kinds of LA that one can only see while riding in a bus.
I have not been to many FOSS meetings in the past--preferring to be a mailing list person. However, one thing that one notices is that there is surge of companies coming to these meetings nowadays. The technical content of the talks was also rather low--perhaps this why it was called "Expo" and not "DevConf". However, most of the talks which I attended were well-prepared and useful.
All in all a great experience. Since it was Kat who pointed out in her posting to TAG that "scale5x" was happening just in my temporary neighbourhood, a big thank you to her as well.
[ Thread continues here (2 messages/2.69kB) ]
Talkback: Discuss this article with The Answer Gang
Kapil Hari Paranjape has been a ``hack''-er since his punch-card days.
Specifically, this means that he has never written a ``real'' program.
He has merely tinkered with programs written by others. After playing
with Minix in 1990-91 he thought of writing his first program---a
``genuine'' *nix kernel for the x86 class of machines. Luckily for him a
certain L. Torvalds got there first---thereby saving him the trouble
(once again) of actually writing code. In eternal gratitude he has spent
a lot of time tinkering with and promoting Linux and GNU since those
days---much to the dismay of many around him who think he should
concentrate on mathematical research---which is his paying job. The
interplay between actual running programs, what can be computed in
principle and what can be shown to exist continues to fascinate him.
