News from the front
By Mike Shaver, 27-Jul-1999
At the time of this writing, the dust has settled around M8, which puts us almost halfway through the milestone list for SeaMonkey (a.k.a Mozilla 5.0). The Mozilla developers are progressing well on the new architecture described in Brendan's roadmap, and it's time to take another look at where we are. The roadmap has stood up well as a description of our technical flight plan; this document should help plot our current position in organizational terms.
Additionally, there has been a fair amount of talk in the press (print and pixel), slashdot and the like about the state and health of the project, so with the help of some new statistics I'll share my thoughts on those topics.
A word about terminology
Throughout this document, the term external will be used to describe someone who isn't paid by Netscape or AOL. It's a perhaps-unfortunate term, because all Mozilla developers are equally "internal" to the project, but it's well-understood shorthand for "not paid by Netscape or AOL" and no other terms recommend themselves immediately as clearer and simpler. It's true that there are Mozilla architecture meetings and the like hosted on the Netscape campus; Mozilla's pre-history and the number of Netscape engineers working on Mozilla makes it more practical to do so. Anyone interested in attending is welcome to, of course, either in person or by telephone. This is true whether you're paid by Netscape or not: I live and work thousands of kilometres from the Netscape campus, so I rely on telephone or just read the minutes afterwards.
Legion of Mozilla
Grovelling through the bonsai
database produces some not-entirely-surprising statistics. For
the month of June, there were 125 different people checking code
in, of whom 108 had email addresses ending in
mozilla.org, and they
checked in 53 patches of varying sizes on behalf
of people without direct CVS access
The M7 milestone and month of June brought us some encouraging numbers about the Mozilla community in addition to the developers:
- Unique posters in the mozilla newsgroups and mailing lists is up to 1049 in the month of June, and we've been averaging 990 a month since January. (Tragically, that number includes a small group of spammers.)
- Those thousand-or-so posters generated 6215 posts in June, up more than 50% from the January total of 4017.
- When they're not posting to the newsgroups, lots of people are giving the software a spin: the four weeks of M7 saw 52,199 downloads of pre-built binaries for Windows, Macintosh, Linux and OS/2 from ftp.mozilla.org alone (statistics from our 100+ mirrors aren't readily available). In addition, the M7 source was pulled from our FTP site 2022 times.
- The bug reporting and fixing rate for June was good, too: in addition to the efforts of Netscape's QA team, 584 bugs were reported by "external" testers.
- B10K: we hit 10000 bug reports and enhancement requests in bugzilla in June, including those migrated from the internal Netscape bug database. (I should point out that the 10000-plus number counts all reports filed in bugzilla, including resolved bugs, enhancement requests and ``virtual bugs'' used for dependency tracking.)
In addition, some contributors aren't represented by our stats. As an example, participants in the recent BugAThon 300 were usually making changes to bugs (adding test cases) rather than filing new ones, so they don't show up in our above June-bug count. Their efforts are certainly valuable, and will make many of our bugs easier to fix and verify, so we should work on capturing data about that sort of contribution.
Isn't that a lot of Netscape?
There are indeed a lot of people paid by Netscape/AOL to work on Mozilla, including yours truly. Some people, including a hearty number of journalists, conclude from that fact that Mozilla isn't an open development effort, or that nobody but Netscape cares about the fate of the project. I think those are pretty silly conclusions to draw, and here's why:
If Netscape were to take all but 20 of those developers off of Mozilla, external developers would outnumber internal by a fair amount. But would that really improve the openness or health of the project? Would it mean that more people outside Netscape cared about Mozilla? Of course not; in fact, the loss of 80-odd developers would set Mozilla back, as would be the case with most open source projects.
So if having a smaller number of Netscape developers wouldn't make Mozilla a better project, why would their significant development contribution make it a worse one? I think it's pretty clear that the ``internal to external'' ratio isn't a useful metric for determining how much interest the open source community has in Mozilla, or for gauging the health of the project.
What do those numbers mean?
One of the hard parts of interpreting these numbers is figuring out how to quantify contributions: a "checkin" can represent a pretty wide range of effort, as can a "bug report". (Regardless of the width of that spectrum, we're obviously pleased to have the help of anyone who wants to contribute.)
Another hard part is that some of our numbers are incomplete: it's hard to generate good statistics for things like CVS checkouts/updates, and we don't have ready access to logs for our FTP mirror sites. (Looks like we had 19158 fresh CVS checkouts in June, but the CVS logging facilities make it pretty much impossible to count the number of updates, or even the number of unique hosts.)
All that hedging aside, and considering rougher historical numbers not included in this document, the general trend is toward larger numbers of developers, increased bug reporting and fixing rates, and ever-growing downloads. (As an example, we're not through July yet, but there were already 5 new external owners added so far this month.) I think those are all good things for the health of the project, personally.
I also think that looking at the rates of change will be
interesting, so I'm looking forward to putting together the
numbers for July/M9 and comparing. Also, I've been mostly
unsuccessful at finding examples of similar metrics for other
open-source projects. If anyone reading this knows of such a
thing, I'd love to
it. The Mozilla project at large -- contributors, users,
*@mozilla.org infrastructure folk and other
interested parties -- should maybe have a
about which numbers matter to us most.
indirect external checkins refer to
patches that were committed to the tree, but which were developed
by a contributor without CVS access. This figure is virtually
always going to be inaccurately low, because it relies on pattern
matching for email addresses in the CVS commit messages, and some
contributions make it in unattributed (oops!) or attributed by
name rather than email address.
 Of the 584 bugs reported in June, 245 were
INVALID, leaving 339 to be
perhaps reclassified in the future.