Highlights

• Chandler M5 due Sept 7. Down to 332 bugs in 0.6. Merged Set/Collection branch into trunk successfully.

Application

• Progress: Branch merge done
• Plans: m5 bugs (regressions caused by Set/Collection work)

Services, Dev Platform

• Progress: I18n plan evolution and implementation. Certificate handling, sharing, timezone and Set fixes. Doc generation work. Performance improvements.
• Plans: Twisted reactor changes and code review. Bug fixing. More performance work. Finish recurrences.

QA, Build

• Progress: Tested Set branch and checkpoint build after merge. Added performance and acceptance test cases to CATS. Bonsai/SVN integration proceeding. Converted Zanshin to FSFS. External makefile now non-recursive. Continuous builds for Cosmo.
• Plans: Extend CATS for use by developers. More work on external makefile. Proposal for instrumentation in Chandler. Retriage build bugs to account for m6. Refactor epydoc generation (hardhat and cron jobs).

Cosmo, Scooby

• Progress: Tomcat upgrade; milestone 0.2-4 (nearing completion and release of 0.2)
• Plans: iCal interoperability, Scooby week view developments.

IT

• Progress: Server for Cosmo demo installed at ISC, running, and basically available. First real SSL cert purchased and installed.
• Plans: New Silk-dedicated host for QA. More SSL certs. Some work delayed while Chris is gone.
• Personnel: Chris away Sept 2-19

Community

• Plans: IRC next week: performance and profiling. CalConnect roundtable and interop testing hosted by OSAF Sept 13-15. Westwood Advisory Council meeting Sept 22 in Pittsburgh (Lisa, Sheila, Mitch).

I attended the FLOSS Extreme Usability Spring II in San Francisco last week.

In the words of the event organizers Blue Oxen and Aspiration:

“Extreme usability is a methodology that incorporates usability in a highly iterative and agile development process and that partners usability practitioners with programmers and users as co-designers and co-developers.”

Unfortunately, we were unable to take Chandler there as a project, but I had a great time working with the folks from CivicCRM. The event lived up to it’s name and was an extremely intense 3 days of whirlwind usability.

I’ve written up a short-ish report of the what our group accomplished at the sprint and the process we “invented” along the way.

The write-up in turn generated some thoughts as to the nature of usability and design, both in general and in practice. In particular, it got me thinking about how the vague and intangible nature of designing software for human beings is better suited to Extremely Rapid and Extremely Iterative design methods than the Somewhat Rapid and Somewhat Iterative approach that I’m used to.

Personal take on What is Extreme Usability?

Broader implications on What is Usability?

I also realized that while here at OSAF, we are priveleged to have such a close and ongoing working relationship between Development and Design, there are still many lessons to be learned about how to make that relationship even better.

I’m looking forward to trying out some of the techniques we experimented with at the sprint, in-house. Perhaps to iron out some of the existing design issues we have as well as to generate new ideas for areas of the UI that are still murky at best. I’d like to gain a better understanding of where Extreme Usability works very well and where it works, not so well.

Highlights

• Chandler: To merge or not to merge, that is the question [for the branch]. We won’t try for the m5 milestone until the branch is merged and return to QA acceptability.

Design

• Progress: Demo and status for CSG meeting.
• Plans: Floss Usability Sprint (Mimi OOF M/T/W. Friday: presentation for California Digital Library. Extensibility discussions. A ton of 0.7 planning tasks.

Applications:

• Progress : Progress on all m5 tasks. Branch is now only a couple of bugs away of being merged.
• Plan : Merge branch in trunk. Finish all m5 tasks.
• Problems : m5 delayed as long as we don’t have the branch and trunk merged. Performance went really down with recent addition. Need to nail why before 0.6 out.

Services and Dev Platform

• Progress: Content model cleanup (in code). API cleanup work carried to sample parcels. Revision to i18n string translation model. Timezone bug fixes.
• Plans: Continue working on set/collection branch integration. Continue content model cleanup to wiki. More resources on performance. Timezone bug fixes.

QA/Build

• Progress: CATS rearchitecture. New perf test cases. Tested recurrence and timezone feature work. ICU/PyICU upgraded. Zanshin revved.
• Plans: Branch acceptability testing. Simplify CATS setup script. Upgrade libxml2.

Cosmo/Scooby

• Progress: Facelift to Cosmo GUI
• Plans: Apple iCal interoperability

• Progress: Cosmo demo server hardware built and awaiting install at colo this week. IT app dev server now in use.
• Plans: Real SSL certs. Server backups improved. Some work delayed while Chris is gone.
• Personnel: Chris Haumesser on vacation Sept 2-20

Community

• Progress: CSG/WAC recalibration meeting past and tentatively successful — communication definitely happened
• Plans: Debrief from the recalibration meeting. More planning for CalConnect meeting Sept 13-15. Start preparing for next CSG meeting Sept 21-23. IRC topic this week: i18n.

I thought I had better post my belated impressions of OSCON 2005 soon or I might as well wait until next year.

This was my first trip OSCON and in addition to simply attending, I was pleased to be able to participate in the CALDAV panel. It was very well attended, with standing-room only, demonstrating that there is clearly a great deal of interest around calendaring and interoperability. Due to some technical difficulties, I was the only one able to get a demo up and running and showed off a few sharing scenarios between Chandler and Cosmo which seemed to generate some buzz from the audience. The BOF immediately afterwards meant more time for Q&A where people asked some tough questions about scheduling. I realized how wide reaching calendaring pain goes when one individual spoke up about the specific challenges of managing calendars for booking a concert venue. The only other related session, was a discussion on the complexity of calendar data interchange, the various formats and clients available. Chandler made the list and did pretty well in a comparison of different clients handling of all day events.

In addition to attending the Chandler presentations and BOFs I also registered for the Open Source Business Review which was a special track of panels geared towards the needs of enterprise businesses. Some of these sessions were more informative/relevant than others. In general, I tried to attend a variety of sessions that I thought would be interesting. I was a bit disappointed that there weren’t more sessions geared towards community building or strategies for product adoption.

A few interesting talks worth noting…

Computational Origami by Robert Lang was extremely fascinating. In addition to impressing us with a detailed cuckoo clock made out of a single sheet of paper, he discussed the math and algorithms behind origami and how those can be applied to solve real engineering problems.

Saul Griffith, one of the keynotes, spoke about Howtoons, one-page cartoons showing kids how to build things with ordinary stuff (Turkey Baster Flute, Shockwave Air Cannon). It was an interesting perspective on how we can expose the younger generation to open-source philosophies and collaboration.

Women in Open Source was a also a great panel discussion. I was surprised to learn that less than 2% of the open source community is female which is significantly less than for closed source development. Based on the animated discussion as well as attendance, I suspect this will be a topic covered at future conferences.

Overall, it was educational, fun and I came away with a much better sense of the wide variety of open source projects that are in the works.

The milestone 3 release of Cosmo 0.2 is ready for download.

Milestone 3 includes a partial CalDAV implementation, including MKCALENDAR, GET and PUT for events (including timezones and alarms), and the new properties specified by CalDAV. Other added features include strong ETag support and full PROPFIND and PROPPATCH implementations.

0.2 is scheduled for release on 09/09. The majority of the remaining work is testing, bug fixes and documentation.

More information about Cosmo is available at the project wiki.

OSCON 2005 was my first conference since I entered the open source world by joining OSAF in January, so it was really great for me to get a sense of the huge variety of open source development that’s going on in the world. In a way, for me the conference was like getting a compressed three-day college degree in open source. Some of the “classes” had little to do with software per se, but were quite fascinating. Examples that stand out here were the presentation on UCSF’s archive of internal tobacco industry documents made public by the industry’s settlement of lawsuits with several states, and Robert Lang’s keynote on the open-source-like approach that has led to amazing developments in origami in the last fifty years, or).

I did attend the OSAF presentations (Ted ’s Chandler parcel session, and the CalDAV panel and BOF). It was clear that there is a high level of interest both in Chandler and interoperable calendaring. In the case of the latter, there were quite a few questions about scheduling (a separate draft from the base CalDAV spec). In general, this is something both technical and non-technical users really want from calendaring, so ironing this stuff out will be pretty important for both Chandler and CalDAV.

Overall, I tried to attend sessions from a variety of different tracks. I the course of my hopping around, there were a couple of session that stood out for me as interesting sagas of real-world deployment and scaling issues, and how systems become complex in unanticipated ways: Brad Fitzpatrick’s talk on LiveJournal’s Backend, and Randal Schwartz’s on anti-spam measures he’s tried over the years on his domain’s mail server. (The latest, a trick with MX DNS records, seems to be working quite well … for now!) Also, Anthony Baxter’s overview of shtoom (his Twisted-based VOIP implementation) was delivered in inimitable style, and gave good insight into implementing telephony standards in today’s internet.

So far as technology goes, I found Johannes Ernst’s presentation on LID pretty interesting (and Dick Hardt’s earlier keynote on Identity 2.0 was an entertaining and accessible introduction to the general problem of identity on the web). While there are competing initiatives in this space, but eventually this is something Chandler will have to think about. There was enough detail in the Ruby On Rails keynote for me to get a sense of what RoR is all about, even though I know next to nothing about Ruby and don’t work on web apps! One catch-phrase that stood out was “xml sit-ups”: the chore of having to edit and maintain XML configuration files (which RoR helps you avoid). In the case of Chandler, too, the move away from parcel xml will mean less ab work for parcel developers!

Highlights

• Chandler 0.6 M5 planned for next week but evaluating delayed projects.
• Cosmo 0.2-3 planned for this week.
• Scooby release 0.1 planning.

Design

• Progress: Second draft of 0.7 tenets and themes and project break-down. WAC prep on deliverables and mobile device research. M5 bug review. Bringing Matt up to speed on design.
• Plans: Dashboard design presentation prep. Prepare for Floss Usability Sprint. Attribute manager GUI proposal.

Applications

• Progress : ItemCollection conversion to sets completed. Progress on m5 tasks.
• Plan : Merge branch in trunk. Finish all m5 tasks.
• Problems : Merge of sets work in trunk is late and will likely delay m5.

Services

• Progress: Set/Collection integration work (e.g. Sharing). Timezones work.
• Plans: ACL client library code

Dev Platform

• Progress: More parcel.xml gone away. Repository and Set/Collection work proceeding in branch. First round i18n refactoring checked in.
• Plans: Clean up Content model Wiki pages. API documentation. Finish eliminating parcel.xml. More Set/Collection work.

QA/ Build

• Progress: Tested checkpoint build 0815. Add perf test cases to CATS. QA intern presentation at all-entity meeting. Bonsai/SVN work. Created FC4 Test server with Japanese locale.
• Plans: Add setup scripts and config files to CATS. Test recurrence and timezone features.

Cosmo/Scooby

• Progress: Cosmo now stores events as specialized JCR node trees — ready for REPORT implementation. Upgraded Jackrabbit (strong ETags and PROPPATCH). CalDAV properties supported. Developed feature plan for Scooby 0.1. Selected Prototype AJAX/JavaScript library.
• Plans: Release Cosmo 0.2-3 (milestone 3) this week. Begin Apple iCal interoperability work. Break down tasks for Scooby 0.1 features. Prepare to present status/plans for WAC.
• Personnel: Interview designers.

IT

• Progress: Asset inventory collated. Cosmo hosting planning.
• Plans: Finish IT app/dev project. Complete system inventory.
• Personnel: Jared began full-time.

Community

• Plans: CSG /WAC Recalibration meeting coming up next week — Pieter, Lisa, Sheila preparing. CalConnect meeting coming up Sep 13-15 also hosted by OSAF. Mimi presenting to CDL and participating in Floss Usability Sprint.

Having interns around always reminds me of those times were, as an intern myself, I had to write reports and how bad at it I could be. I had one strike of good luck though in the person of my mentor Georges Bronner who gave me a terrific 15 minutes primer on technical writing I will never forget (and that was 20 years ago folks…).

I won’t say I never sinned again and committed some terrible documents (have mercy…) but this short set of advice always worked wonder whenever I applied them diligently.

Talking about it with Katie, she encouraged me to write them down. So, for everyone’s enjoyment, here is Georges’ Sieve written down at last.

Thanks Georges!

Highlights

• Chandler M5 milestone will occur either on Aug 24 or when Item Collection and Set work is checked in. A lot of Chandler development progress evidently being made right now.
• Scooby work officially kicked off with Matt Eernisse’s arrival.

Design

• Progress: Prepared candidate themes or tenets for work in 0.7 release, as well as SWAGs for high-level projects.
• Plans: Finish virtuality paper, work on PIM workflows presentation

Application

• Plan : Alec helping John more closely on Sets and Collections. m5 tasks.
• Progress : Lots of people at OSCon last week. Some progress against m5.
• Problems : Sets and Collections a risk, likely making m5 slip
• Personnel: Last week of Arel Cordero’s internship!

Services

• Progress: Some SSL bugs fixed. Timezone support on individual events. Lots of sharing progress. Zanshin code reviewed.
• Plans: Checkin and review timezone support. Recurrence proxies.
• Problems: ACL support in Zanshin at risk.
• Personnel: Jeffrey Harris to start full-time, in-office, Aug 29.

Dev Platform

• Progress: Lots of progress in schema API and replacing parcel.xml. Logging framework cleaned up. Lots of i18n progress.
• Plans: Content model review. Vote on getting rid of parcel XML. Set integration work.
• Personnel: Andi on vacation.

QA/Build

• Progress: Tinderbox SVN integration complete. Multiple build bugs fixed.
• Plans: Finish Bonsai SVN integration which is nearly complete. Fix bug 2818. Launch/host more automated tests (CATS) making use of CPIA Script.
• Personnel: Last week of Anthony De Franco’s internship!

Cosmo/Scooby

• Progress: Parsing iCalendar bodies in PUT requests. Integrated Jackrabbit upgrades. OSCon panel. Evaluated AJAX libraries for use in Scooby — tentative recommendation to use prototype.
• Plans: Evolve Brendan’s mockup into a data-backed WebApp — focus on dynamic and sexy calendar week view.
• Personnel: Matt Eernisse joined team last week and is spending this week in the office.

• Progress: Plans to support Chandler 0.6 testing and usage by hosting Cosmo. Server database active.
• Plans: Complete application development infrastructure work for IT projects. Complete systems inventory.
• Personnel: Jared Rhine is now full-time with OSAF

Community

• Progress: OSCon panels and presentations complete.
• Plans: CSG Recalibration meeting coming up Aug 22-23 — Pieter, Lisa, Sheila preparing. CalConnect meeting coming up Sep 13-15 also hosted by OSAF.

I popped up to OSCon last week for the Wed afternoon CalDAV panel and BOF. I was a little bit surprised and not a little pleased by the standing-room-only attendance at the panel. Things went very well considering the nonexistent amount of detailed planning we’d done - Pieter Hartsook is owed thanks for saving our butts by putting together an agenda and giving us a bit of direction

Observations from the panel:

• I was the only server guy to actually discuss my product’s feature set beyond a general statement of CalDAV support. Unfortunately, because we had not led off the panel with a discussion of what CalDAV actually is, I think some of it went over peoples’ heads :). Still, given the questions I got during the panel and BOF, I’m pretty sure most people got most of what Cosmo has to offer.
• Pretty much all of the other server teams are approaching their products as web applications (presumably on top of SQL databases) that speak CalDAV out the side, whereas we at OSAF are building the protocol server and web application separately, using CalDAV to communicate between the two. You’ll be able to aim Scooby at the UW Calendar and Hula servers, but it’s not clear if you’ll be able to aim one of these servers or WebCalendar at Cosmo.
• In general, the problems that people are having are more complex than the ones we implementers are able to solve at the moment. We’re still getting our feet under us, just becoming able to get and put events interoperably between multiple clients and servers, but the audience asked hard questions about enterprise integration, complex calendaring problems and and advanced scheduling features that we’re very far from being able to address. We’re happy just to have gotten this far (it took ten years to get a calendar access protocol that anybody actually wrote code for), but our users need us to get a lot farther real fast.
• Calendar web applications are popular. Seems like everybody’s working on one. It’s going to be interesting to see how the multiple projects differentiate themselves.

I’m looking forward to ApacheCon, where I’ll hopefully have an hour to explain a lot more about CalDAV and about Cosmo, including why it’s different and more interesting than the other guys’ servers