bzr: bzr+ssh works, much like svn+ssh, w/HTTP option for public repositories. You can start your own private branches, and if you have permissions, you can ‘push’ back to the trunk. Common commands: ignore status log whoami init add branch pull merge push diff commit blame bundle gannotate pops up GTK+ window to let you annotate (there’s a pluggable arch)

Same session, (Brenton LeanHardt) is going to argue that you should only use a distributed VCS for a multideveloper project, on the grounds that you have more flexibility of working your VCS into your workflow rather than the other way around. SVN requires you to remember a magic number to merge (yes, yes it does). This stops people from committing their changes.

Q. How do you take the cacophony of private branches and work out a coherent release?

A. It boils down as you go up the network of trust; small groups of developers can tolerate failing tests, crazy experiments, etc., but as team leads ship code upwards to testers and release team, each subsequent level trusts that the process and people at the previous level will get it working (there’s still a centralized line of development). So roughly, IIUC, you localize the cacophony.

Somebody in the audience mentioned “Streamed Lines” a paper on managing multiple lines of development in version control systems.

12AM: caterer is late, soo …Clinton R. Nixon: Test Driven Development. Show of hands; most folks are using xUnit or similar.Non-xUnit: TJ Stankus mentions AutoTest (continuously runs Ruby-specific) (available in gem: zentest). Testing webapps: Selenium; watir can simulate clicks in the browser, but only supports IE. Also windmill. Using tests for documentation: e.g. name test case thisShouldThat; also see RSpec.

2:30PM: Jason Rudolph talking about Grails ; RoR-inspired Groovy web development “convention over configuration”,
“sensible defaults, opinionated software”; Groovy because http://www.jasonrudolph.com/downloads/presentations/Getting_Started_with_Grails.pdf ts integration with
Java is two-way (extend Java classes in Groovy and vice-versa). 1.0 in 2007.
Hibernate, Spring, Quartz are all integrated; everything is very Rails-like,
routing tables, mapping of URIs to controller classes and methods. Changes
don’t require explicit recompiles in most cases, including i18n files (!)
Also, dynamic generation of DAO-style methods (
Customer.findByState('NC') )

3:30PM: behold the OLPC:

Greg DeKoenigsberg showed off the OLPC; interesting facts are that it’s not a “hundred dollar laptop,” and that its most expensive component is the keyboard, because it had to be developed for the project while every other component is off-the-shelf.

4:30PM: Jason Rudolph is leading a session for folks to share Mac
productivity tips. [what follows is mostly for my memory, it's pretty
telegraphic, sorry] Quicksilver. Command-H (hide window); use a script to
“clean” your desktop (drop down to one app quickly). TextMate SVN integration
(Ctrl-Shift-A). VirtueDesktops for multiple desktops (which won’t be needed in
Leopard). Sleep with command-option-eject. Pathfinder instead of Finder, e.g.
instant-on terminal in currently selected directory. Growl (notification
framework). iTerm to replace terminal (also should be improved in Leopard).
TextExpander (cross-app abbreviations); can also pull contents of clipboard as
parameter to the expansion macro.

OK, enough rambling.

Intros. BarCamps explained, the guiding principle is informality and fluidty (wandering around the sessions is encouraged, etc.)

“Keynote” is over, people who want to propose a session are gathering to make their pitches.

The “scrum” is about to begin, where the sesions are going to shake out. I’ve got three I think I want to go to (APP, CI, and the busines of open source)

(ok, not liveblogging exactly because networking hasn’t been easy)

Morning sessions

learned a few things about how apps are delivered to mobile devices (aka “cellphones’), including testing (how do you do it with so many different clients out there?)

Seairth Jackson Jacobs (hope I remembered the name correctly) talked about RNA, RESTful Notification Architecture, which is a callback-based messaging protocol over HTTP. The basic idea is that participants in the conversation first exchange messages to eastablish URIs where further messages can be sent. The basic idea seems pretty solid, and one of the things it does is pass URIs to message contents instead of the message contents themselves. We spoke a little afterwards, I think this is promising.

Continuous integration: we spoke of CruiseControl and Continuum, and were walked through some basic demos. What more is there to say? Well, on my part this much, I guess.

Atom Publishing Protocol: the donuts of content management (is there anything it can’t do?) Whatever one may think of the wrinkles, it’s a well thought-through protocol.

The future of publishing: the general direction this one went in was that, at least with respect to books about the computer industry, getting a book published is going to look more and more like iterative development of software, with ‘beta releases’ put out there for public comment (the Pragmatic Bookshelf is already doing this).

Agile development: lots more here, with folks who are walking the walk trying to provide guidance. A lot of the time was focused on how you square agile software development with QA and requirements (which are, in a certain sense, what sandwich the traditional development cycle). Strategies for dealing with impedance mismatches between how the clients expect to generate requirements and what developers need to meet those requirements, and also between pervasive automated testing and the traditional role of software QA.

update:  if you’re wondering why there’s a trackback to Ryan Daigle’s “BarCampRDU reclamation project” from this post, it’s because this would be my collection of notes.