The Goldfish Chronicles

From the members meeting today:

       Stats (please feel free to share!):
         Committers: 2359
         Members: 298
         iCLAs on file: 3583
         Projects: 84
         Incubating: 36
         Labs: 30
         Attic: 8
         SVN Revisions: r962795
       Web stats:
         Jun '10 average successful requests/day:
           www.us: 6,014,887  ; www.eu:    37,954
           svn.us: 2,091,040  ; svn.eu:   985,331
           issues:       N/A  ; people:   151,896
         Jun '10 average data transfer/day:
           www.us: 654.99 GB  ; www.eu:  94.42 GB
           svn.us:  30.72 GB  ; svn.eu:   5.99 GB
           issues:       N/A  ; people: 108.11 GB
         See: http://people.apache.org/~henkp/analog/
         (issues machine was compromised per incident noted above)
       Mail stats:
         Connections (June 25, 2010):
           nike (mx.eu): 302,417 (97.63% rej.);
           athena (mx.us): 321,519 (97.66% rej.)
         Most useful plugins:
           dnsbl (70%), require_resolvable_fromhost (21%)
       Mailing lists and SVN commits graph:
         http://www.apache.org/dev/stats/

[Publishing a draft from a few months back. Internet time is lovely - in the meantime SourceForge bought Ohloh and not Google as I predicted. Given that I was predicting it would be bought by Google Code as such, I’m feeling nicely prescient here]

How the Open Space became

Nathan Yandell - Year 12 - OSS101

In the last decade, the 10s, Open Source development has undergone yet another revolution.  The bazaar has truly come of age and the last Open Source cathedrals are torn down - there are no more ‘committers’.  Looking back, it’s easy to see how we got here - the only oddity is why it took so long.

The first step was the surprising popularity of Linus Torvalds’ Git project.  Written for the now gone Linux operating system, it took off, added recognition to existing tools and was emulated by others.  It’s unlikely to cause argument to say that without Git we would not have had dSVN.

The second step came when Google having bought Ohloh, decided to donate the company to the Open Source Initiative and signed Google Code up as the first user of its shared identity pool. Sourceforge and Codeplex soon followed - eventually even the hold outs at the FSF, Mozilla, Eclipse and Apache joined the pool.

Sourceforge’s ‘branch a non-sourceforge project’ feature made a high noise but low impact entrance - but over time it built in popularity and with the other open cathedrals following suit the Open Space was born allowing projects to be copied from one repository to another.

To an outsider - our new world looked like chaos. URLs to codebases were as fleeting as your parent’s tweets. Reputations were built up and torn down, code patches danced into view only to vanish into history. More worryingly, legal arguments between projects increased. We saw amendments to the popular Open Source licenses that defined the list of licenses which may be used on a patch branch, and trademark enforcement was used to keep terrible old software from being destroyed by this week’s new up and coming patch branch.

This all changed with the release of the Creative Commons Trademark License (Swaziland). Connected to changes in trademark law in Swaziland, it allowed for trademarked brands while enforcing that product names must be openly useable. A dusty project such as Firefox saw a name change to Mozilla Iceweazel - at long last able to build on the success of its descendants. GNU Java finally gained competitors in IBM Java and Codehaus Java, which led to much confusion in the aging Java development community.

Now I can take a project, fork it and apply my official IETF Brand name by signing the product’s Ohloh certificate with my personal Ohloh brand certificate. As a member of the OpenLoonies I can also apply the certificate for the OpenLoonie brand (”The most trusted brand in Open Source”) - once I’ve had 3 +1s from other loonies. Within an hour I can release my bugfix as OpenLoonie Lucene and it can start building up its trust rating.

How people used software before the community evolved the trusted open software system - my generation will never understand.  It must have been like flying your car blind.

Roller JIRA was also moved into the main ASF JIRA. Not a bad vacation - three JIRAs turned off Also saw the last stages of the Commons CLI 1.2 release being done.

Cayenne join Click on the central JIRA. The machine flexes its new free memory. The admin looks forward to the ease of only having 1 instance to manage someday.

Next up - Roller.

Step 1 in the 5 step plan to merge your JIRAs is complete. The Apache Click JIRA has now been merged into the main JIRA and the underlying machine has received a nice memory refund from the JIRA world.

Next up - Cayenne JIRA.

Happiness is a successful upgrade.

3.13.2 is an interesting JIRA upgrade for us. Major new features:

#1# We can start merging the JIRAs and saving some cpu/memory/disk. Most importantly saving admin time.

#2# Favourite filters. Atlassian have changed filtering to be less spammy by having you star the filters you like. I’ll have to adjust my plugins on the front page to adapt to this, but it’s definitely a good step on their part.

#3# Sharable dashboards. HUGE new feature. Now I can put together a Commons dashboard, with perhaps a custom plugin just for Commons, and everyone else in Commons can share it.

Most importantly - this was done before ApacheCon rather than taking JIRA down for everyone while there, and before I am on vacation freeing me up to do things that are more original with my evenings at my parents. Or freeing me up to look into merging the JIRAs

Upgrade happening Sunday 21:00->23:00 PST (Monday 04:00->06:00 GMT) to JIRA 3.13.2 - see timeanddate.com for more times.

JIRA will be read-only for the time period, and down for a short time therein just before it becomes writeable again. Please raise questions on irc.freenode.net #asfinfra

[The other JIRAs were migrated on Friday night - Struts, ActiveMQ/ServiceMix, Cayenne, Click, Roller]

Deployed the new Maven2 based Taglibs site:

Old site screenshot:   http://people.apache.org/~bayard/OldTaglibsSite.png

New site screenshot:  http://people.apache.org/~bayard/NewTaglibsSite.png

Not much change - which is the intent. More of a change for the underlying taglib pages themselves - here’s a sneak peek.

I’ve made more commits (96) to the Apache repository in the last 4 days than I did in any month last year. 129 commits in July 2007 is the last time I hit such ‘heights’, which was when Commons moved out of Jakarta.

So what have I been doing? A few things. Nothing exciting, all the kind of stuff that I like to plod along doing. In Jakarta land I’ve been migrating the taglibs to Maven2. I started it a little under a year ago and lost steam. Tried it again on the 3rd of the year and it went well. So the last couple of days have seen more activity in Taglibs since… *Vader pause*.  Major goals are: 1) Make it easier for newcomers. 2) Make it easier to test by figuring out how to integrate Cactus into a parent pom. 3) Link the energy to the forthcoming JSTL 2.0 release. Major bit I’m dreading is running the JSTL 2.0 TCK.

What else? A bit of gump committing. When you start ripping around with components, you break the gump build. Fun. Also an email to Cactus to start learning about Cactus+Maven2.

Another is the Apache Attic site. I got stuck trying to copy the main Apache setup over. Finally commited it on the 1st in the hope someone else would solve it for me, then with the fresh energy on the 3rd beating up Taglibs I realized my mistake and have a 1 page site sitting in SVN ready to be deployed when enough time has passed for people to yell at it. The Attic is a place to (somewhat) put the projects that have gone inactive. Ironic that while I’m whittling away at that, I’m also providing life to projects that are not too many steps away.

Lastly - resolving the only open bug in the Commons Collections 3.3 release list. Technically the work was done on the 31st, but the commit was done today. Version 3.3 has been ready, for some value of ready, for release for a while but I just don’t have the energy to go through the release process. Not so much the process - that’s just sending an email and waiting for people to find bugs for you. I can’t remember how to do the M2 release steps and am prevaricating.

Welcome to me - VP of Apache Attic despite my spending a year scuffing my feet and hoping someone else would want to wear the hat.

What is Apache Attic going to be? It’s a consultancy project to provide janitorial services to zombies.