Oleg's wiki

HelloWorld

Sun, 27 Jun 2010 00:33:00 GMT

Wiki (n) - A collaborative site where users can edit content and have fun.

Welcome to my web scrapbook. Notable scraps include:

BarcampOxford 2010
TransferSummit 2010
BarCampApacheOxford 2009
QConLondon 2009

If you're bored already, how about a little game of jsTetris - just for fun? (send a msg if you beat the high score of 288933).

Other areas of interest should be floating about in the cloud .. Questions? Comments? Missing tiddlers? Get in touch :)

BreathingEasy

Sat, 26 Jun 2010 15:46:00 GMT

Respiration Coach from VeRSI
https://www.versi.edu.au/versi-projects/life-science-projects/respiration-coach

BarcampOxford

Sat, 26 Jun 2010 15:41:00 GMT

RepRap
EclipseTigerstripe
OpenSource101
ReleaseManagement
OxfordGeekJam
TestingInRubyLand
CodeSense
SemanticApps
BreathingEasy

SemanticApps

Sat, 26 Jun 2010 15:41:00 GMT

Sander @sandervdwaal
http://simal.oss-watch.ac.uk - 1500 GIST funded projects, interfaces with ohloh.net
Main application is to match projects by theme, geographically, finding connections
Ross has public fingerprint, geo info, current projects and contacts in his FOAF.. but does Ross have any friends?
FOAF address book developed by Henry Story, works on foaf+ssl and distributed auth mechanisms
If you have your own FOAF RDF file, it is accessible to anyone but controlled completely by you
To maintain a certain level of privacy, you need to have a way of restricting public access
Lots of recent buzz around Semantic Web, Facebook's Open Graph, Public/Open Data
Archives Hub in Manchester aims to open up and connect lots of public data
Difficult to work in this area - creating the data is easy, but making sure the data is clean/appropriate, building flexible enough UI's
Distributed Flickr - finding people to add to your social circle, get a snippet on my page of everyone I follow
FOAF is not the only way forward.. any alternatives?? Room for competitive approaches
The concept of ontologies is a very useful one
Cool applications: find some data sets and merge them (see Economics talk from this morning..)

CodeSense

Sat, 26 Jun 2010 14:29:00 GMT

Nasir Jamal @_nasj
http://lcn.com (ISP)

Writing better code

Understanding easily when coming back later
Mostly using Ruby
don't always write perfect code, sometimes slip to the dark side

How to measure code quality

it's all relative..
there are different tools available to measure complexity, structure
the only valid measurement of code quality: WTFs per minute

Practical example of a class

244 lines of code .. too many!
collapse and see that just one method has 141 lines
Run is too long, doing too many things
difficult to test, reuse
redundant comments
not "Don't Repeat Yourself" (DRY)
instead of refactoring, people kept adding stuff (or copying the class and changing it)
"Don't Be A D***" (DBAD).. refactor

Teams

Having the courage to take actions
More benefits from general code ownership than from individual blame
Instill a sense for code quality in every team member by leading through example
Code review process? Peer review leads to controversy
Deliver comments in a constructive way - tell people their code is **** but.. be nice
Always have some positive input as well as criticism

Refactoring

(see slides)
metric_fu is a tool for getting automatic reports in Ruby to measure improvements
Perlcritic for Perl, lots of Java and .NET
Biggest aim is an improvement in release/deployment cycles

TestingInRubyLand

Sat, 26 Jun 2010 13:53:00 GMT

TODO: get notes from @timblair http://tim.bla.ir/

OxfordGeekJam

Sat, 26 Jun 2010 13:52:00 GMT

History, motivations, experience
Demo projects
How to run a freestyle collaborative coding event

Questions

marketing the event - what's in a name? define:geek
how to we make sure we don't use up too much time brainstorming
ensure that laptops come out not too early to give enough space for design
subversion repository as a key way to gauge progress and so on
continuation of projects - is everything just throwaway material?
maybe different ways of scheduling would fit better, e.g. every first Thursday or whatever

Tweets/links

http://twitter.com/chrisjdavis/statuses/17084999307
http://www.oxfordgeekjam.net

ReleaseManagement

Sat, 26 Jun 2010 13:49:00 GMT

Hyrun Wright
Working on release management as a PhD topic and looking for people to interview

Release history of Subversion

Maintained svn for years
People don't talk across organizations, lessons learned are hard to apply
Little best-practice knowledge in most projects - how can we learn from each other?
Internal release testing process

Long time between releases

Shoehorn too much stuff into the release
Testing and validation took a lot longer than expected
The bigger the product, the more bugs - and developers don't want to shift bugs
But at some point, you've got to ask if it's worth it ...
The design of merge tracking evolved during 1.5, in a non-modular way with major scope creep making the complexity involved (and there was a massive bottleneck because of the specialist knowledge required to get near the feature)
Takes a lot of discipline on the development team (few developers are willing to back out on work to get a release out)
Disable access to half-baked features which most users are not going to get to
What are developers used to, talk to people about their preferred conventions

Perspectives

Even a small company has to be very careful about each release
Platforms make release processes important if other development depends on you
Releases are a labor intensive process
BUT release process is (usually) not the product differentiator
Large shops that deploy SVN are happy with once a year upgrades - more conservative release schedule
Regression testing, test driven development, agile - the usual suspects
Policies of releasing unfinished code to an appropriate channel
Is it a problem of accurately estimating the time it takes to do the work (very common, very difficult problem)

Balancing the Interest

Sometimes it can be a real battle to get a project out the door
Convincing the other side that it has values is hard (meeting the needs of many actors)
Keeping track of the process and issue
Breaking features up in the right way: when we tried to ship SVN 1.5, we couldn't until a particular feature was dones
Certain cost to creating releases to track of: dev teams shut down for 2-3 days, testing and reviewing, customers have to upgrade the platform
Find someone with thick skin who understands the situation, you can trust and let them break the alpha copies
Tell people to stop making hard core changes in trunk
Needs dedicated resources to manage releases (and commonly these are business and not technical people that have 2nd-degree experience of the challenge)
Tried time-based release cycles - calendar of fixed periods for different release types (fails once you get past the low hanging fruit and there is a degree of uncertainty)
Ticketing/tracking systems really help to keep balancing the interest

Questions

When do we make the release?
What type of release is it (bug fix, feature)?
Mechanics of distribution involved?
Keeping services running smoothly through upgrades?
Product or platform? (e.g. Subversion is both)
Trunk or branch?

Tweets/links

http://twitter.com/hyrumwright/statuses/17089331531
http://twitter.com/adrianh/statuses/17091330681

RepRap

Sat, 26 Jun 2010 13:49:00 GMT

RT @ni: Ooo... they have a RepRap set up at BarcampOxford :) http://reprap.org/wiki/Main_Page #barcampoxford #repr... http://bit.ly/cSJ9ug
http://twitter.com/barcampoxford/statuses/17084170767
£300-400 of materials
check out http://omg3d.com voodoo digilab voocat
IETF mime type for matter transportation? :)
Humans and machine in a similar symbiosis to plants and animals

OpenSource101

Sat, 26 Jun 2010 11:03:00 GMT

Noirin Shirley
Technical Writer Google

Noirin Shirley is a jack of all trades, and a master of several. A technical writer by day, Noirin loves her job in Google Zurich, and gets to spend 20% ofher time helping out with Open Source projects - small wonder then that she'son the Project Management Committees for the famous httpd and the newerCommunity Development projects at Apache, as well as being a member of theInfrastructure team, and the Vice President responsible for events!

Values

Be polite.. ask if it's OK!
Everything I learned for Open Source, I learned in kindergarden
When you start to play with others, you can build some pretty cool stuff
It's not as simple as slapping an OSS license...

Best practices

Start releasing early and releasing often
Let people see the pieces and reviewing the small patch
Don't succumb to not-invented-here syndrome
Sometimes you need to sit quietly/lurk/watch, the path is not spelled out
Some projects have alot more documentation (how to contribute, etc.), than others
Sometimes the project documentation is out of date
Neat and tidy: leave things better than the way you found them
Don't be afraid to ask for help!

Communicate

the worst thing that happens is a flame? sometimes there's some valuable content .. read between the lines
try not to send out those emails yourself .. critique the code, not the coder.. excersize some self control
where possible, include positive comments - thank people for the contribution
test your email client (to make sure the email goes to the list or individual)... double check !
United Nations: important to remember that we're from different cultures (and what you perceive to be good manners might not be the same for someone else)
admit your mistakes, be up front

Trust

until you get to the step where you're building trust, you can't be part of the community
review other people's contributions (you can do this before you're making contributions yourself!) .. another pair of eyes is always a good thing
try to get high-bandwidth-interaction with other people (meet in the same room, make 1:1 connections)

Theory to practice

Where do you get started? Often it's not a straight paved road and can be a bit of a maze
Have a greeter for the project, and when a new contribution comes in welcome people (common problem for first timers is nothing happens when they patch in) - not necessarily appointing someone, but encourage everyone to take 5 minutes to be nice to new people in the morning...
Code of conduct - warnings if you step out of line (private mailing list for this purpose...), one of the requirements is to initiate new members
Not having a mailing list can be a disaster (e.g. GitHub, where people throw something on or email the author)
At Apache, we can't figure out how to make the cultural walls work on Git, the lists are essential for communicating and keeping up
At Wikipedia, if you don't create a way of DM'ing people will use mailing lists for that purpose - those personal messages will be out in the open
Ross: you don't get a community going by herding cats - if you want to change the process, better to show by example
Nick: it just took one person to write the 15 do's and dont's on an Apache mailing list that helped everyone

What would make open source communities more attractive

Finding a role: specific activities to start contributing - Kiraly spoke at Apachecon last year about including non-bugs in the bugtracker (e.g. logos for website)
Markus: it can be very intimidating to get started, but the community does have to be very accepting of new people - e.g. contributions valued even if they are not immediately used - in larger projects there is an emphasis on 'core code' that requires in-depth knowledge - a simple 'thank you' suffices
Legion of Bastion Castle - no matter how big or small the contribution, they had a credits page with everybody's name
Eclipse Labs - things that are not part of core Eclipse (bits and pieces, non-core plugins and features) are collected on a Google Code project - association with the core project
The author doesn't always want to maintain the patch, just wants the product to work - try to recognize the interest of the contributor
We don't always recognize or remember the fact that not everyone wants or needs to follow the path (there is room for once or twice off contributors, people have different needs)
Start a community development project to help people get past the initial steps

EclipseTigerstripe

Sat, 26 Jun 2010 10:27:00 GMT

Craig Gallen
http://www.eclipse.org/tigerstripe/

Tigerstripe

Model Driven Engineering
Apache License, Eclipse incubator project
Why Tigerstripe?
* Users of MDE frameworks may not be UML experts...
* Enterprise Models are BIG...
* Smaller pieces of Enterprise Models are owned by various Teams...
* MDE involves multiple roles...
Telcos and other industries have very complex models for enterprise software - create an open source way of connecting the different models together
Acquired by Cisco, but we worked together to improve it and widen its use
The model is stored as a combination of descriptive XML and normal Java classes
Start with UML and apply to any target language
Parse the model and generate code
Based on Eclipse, so is completely multiplatform
Most of the work is done internal at Cisco
Potential for wider use
Positioned against EMF - working to align with them
When you have a large enterprise model

OpenOSS

Craig is leading the OpenOSS project (includes Nortel, Sienna, HP, BT, ..)
OpenOSS/J was a forerunner (a Java set of interfaces for Telcos, based on Java 1.4)
http://openoss.sourceforge.net/
http://www.tmforum.org/browse.aspx

TransferSummitLinks

Sat, 26 Jun 2010 09:45:00 GMT

http://www.opencastproject.org/project/matterhorn
http://www.educause.edu/Community/MemDir/Profiles/OlafASchulte/62738
http://www.guykawasaki.com/books/rules.shtml
http://incubator.apache.org/learn/rules-for-revolutionaries.html
http://jroller.com/TedHusted/entry/prim
http://en.wikipedia.org/wiki/Apache_License
http://mail-archives.apache.org/mod_mbox/roller-dev/201005.mbox/browser
http://producingoss.com/
http://www.hawthornlandings.org/2009/03/getting-started-in-open-source-you-dont.html
http://wiki.laptop.org/go/WikiBrowse
http://www.mediawiki.org/wiki/Dedicated_Wikipedia_editor
http://wiki.semantifi.com/index.php/App_Ideas
http://www.liip.ch/
http://carrierdetect.com/
http://www.ch-open.ch/
http://reprap.org/wiki/Main_Page
http://groups.google.com/group/barcamp-oxford/browse_thread/thread/92b540c047c089bd?pli=1
https://mozillalabs.com/raindrop

TransferSummit

Sat, 26 Jun 2010 09:43:00 GMT

TransferSummitTweets
TransferSummitLinks

Day 1

StevenPemberton
AndrewFennell
MarkTaylor
MatthewLangham
SimonPhipps

Day 2

RolandHarwood
MatthiasStuermer
PhilAndrews
JustinErenkrantz
AndrewKatz
SamuelKlein

TransferSummitTweets

Sat, 26 Jun 2010 00:24:00 GMT

http://twitter.com/ShamalFaily/statuses/16917830497
http://www.flickr.com/photos/webmink/4732715052/
http://twitter.com/J_JKMi/statuses/17000531925
http://twitter.com/osswatch/statuses/17021533298
http://jroller.com/TedHusted/entry/prim

SamuelKlein

Sat, 26 Jun 2010 00:23:00 GMT

Learning to scale
http://meta.wikimedia.org/wiki/User:Sj
Wikitime
Simplest way to edit...
500 servers, 200K editors, 6 Gbps
Most traffic is through mirrors and forks
Find network effects, classify
Eliminate bottlenecks, give ownership
Deletion patrol
All the stats are public.. server outages are usually not reported by admins
Share passionately: maximize reach, not margins
Give people a sense of collaboration and participation
Invest in APIs (* not rate limited, corporate accounts boost connection) and models (!)
Varied, friendly core community (!)
WikiBrowse
1.5 million people have an XO http://laptop.org/en/laptop/
Anyone who develops something useful gets one for free contributors@laptop.org
Reproduce nearly every part of the XO through an open design
8yr old girls in Nigeria were running a laptop hospital making their own hardware repairs

WikiBrowse

Fri, 25 Jun 2010 15:29:00 GMT

http://wiki.laptop.org/go/WikiBrowse
Developed by a commercial company
One laptop per child
100'000 families using Wikipedia offline in Peruvian Amazon
Parsers API were available and mwlib was created
Offline editor
OpenMoko Wiki reader (open hardware + software + content!)

JustinErenkrantz

Fri, 25 Jun 2010 11:54:00 GMT

@jerenkrantz
Contributing to httpd since 2001
70+ projects (TLPs)
Apache is a community of developers and users
^ aims to provide a pragmatic non-technical framework to projects

open development* vs open source

^ all development is made in public in a very transparent way
To each his own motivations..support..fun/technical challenge
Not "a patchy web server" ...
Whomever has the best idea leads .. until a better idea is presented
http://incubator.apache.org/learn/rules-for-revolutionaries.html
Earn karma independently in each project
Break patches up into small chunks to make sure they are reviewed (contributor responsibility!)
Code can be vetoed, releases can not - Vetoes... judiciously used
3 votes minimum (ensures diversity & true community)
72 hours minimum time frame for votes (accomodate long weekends, err on caution)
Trolls exist.. don't feed them (and don't become a poisonous person)
ASL2 can be used in GPL3, but no GPL in ASF projects
Umbrellas are bad - agreed on flat organizational structure and spun-off the projects
Once new projects get a mentor they are put into labs - shared mailing list, no non-committers, no releases
http://producingoss.com/
http://www.hawthornlandings.org/2009/03/getting-started-in-open-source-you-dont.html
http://www.bsdcan.org/2009/schedule/events/149.en.html
SCALE_Final_allman_lhawthorn.pdf
http://theapacheway.com/

PhilAndrews

Fri, 25 Jun 2010 11:54:00 GMT

Europe Regional Director, Red Hat
Challenges of open source for commercial use and making OSS enterprise consumable
Do you know what Oracle's operating margins are? (Cost of proprietary ..)
Alternatives to a range of commercial software
Alfresco, INGRES, Jaspersoft, SugarCRM, EnterpriseDB
Very difficult to sell licenses to companies that produce commercial software on one side and adopt open source on the other
EU is very pro-OSS
Drive the OSS market and represent highest value for money
'@J_JKMi: Discussing the delicate balance of keeping the customer, the community and the shareholder happy #red hat #ts10'

MatthiasStuermer

Fri, 25 Jun 2010 10:49:00 GMT

@maemst
started helping Nokia test Linux and open source 10 years ago
good story of how someone from the community proved the Nokia 770 engineers wrong on the flash RAM as cache
Nokia bringing Z500 running Meego to market in 4Q10 http://mynokiablog.com/2010/06/21/arm-based-meego-nokia-tablet-z500-is-out-to-the-world/ ?
Motivations of contributing to open source: Georg von Krogh, S. Haefliger, S.Spaeth, M.Wallin, "Motivations andincentives in open source software development". 6th Annual International Open and User Innovation Workshop. Harvard Business School and Massachusetts Institute of Technology. Boston, August 4-6, 2008
Plus findings concerning FOSS from Nokia:

Low knowledge protection costs
Learning effects
Reputation gain
Adoption and lower costs of innovation
Lower manufacturing costs
Faster time to market

Costs:

Difficulty to differentiate
Loss of business secrets
Giving up control and network entry barriers
Organizational barriers

Interesting way to contrast factors of Control vs. Openness in OSS, plus open source adoption matrix for diverse international and Swiss companies

"This network becomes more important than trade secrets"

- Ari Jaaksi, head of OSS at Nokia
Presentation from 1997 with more background on the Nokia work

RolandHarwood

Fri, 25 Jun 2010 09:31:00 GMT

http://www.nesta.org.uk/
http://www.100open.com/what-we-do/
'There are two ways of making innovation happen: start with what or start with who.' @mberry
'@rolandharwood from Nesta explaining what's difficult about open innovation' ~@SaraAllison
Innovation 'jams' with Virgin, IBM,...
'Roland Harwood: In open innovation you're not buying a product, you're buying a process, a potential...' @jamiedsmith
What happened to http://www.openalchemy.co.uk/page/about/index.html ??
Innovators }RISK INERTIA DISTRUST{ Corporates