Recently, I've started to dive into mozilla with a developer eye. *very
slowly* since my c++ skills are almost non-existant (and my c skills
are, hmmmm, rusted and ruined by the java garbage collector :-)
Anyway, the cultural differences between their style of development and
ours are striking.
1) they have *ONE* big CVS module where all developers can commit to
anything.
I believe this is a *major* mistake because devs have to download the
whole thing in order to build mozilla. the communicator philosophy was
attached to their very mindset. also notice that the mozilla CVS module
is 650 Mb (as of yesterday), this rules out almost all dialup or
pay-per-minute fee users. Which means, in the last 5 years, probably 80%
of the potential non-US developers!!! While DSL is changing this for
europe and other first-world nations, this is not so for the rest of the
world.
with smaller and more manageable CVS modules (and nightly snapshots to
route around CVS firewall restrictions), *our* potential dev pool is
orders of magnitude bigger than theirs.
2) tinkerbox is a nightly build system which aims to improve continuous
integration by forcing everybody to commit to the same tree. I think
this can't scale as much as we can with a Gump-like approach. The
continous forking friction (chimera, phoenix, minotaur) is tearing their
communicator-inflicted mindset apart. Rightly so, IMO. The more diverse
the development community becomes (and this is slowly happening, also,
maybe, because of increase on european broadband), the more the
community will look for 'KISS' solutions that are driven by lazyness and
'getting the job done' by small incremental steps.
In this new mindset, their infrastructure will have to change
significantly to adapt to this.
3) mozilla.org doesn't have the concept of separate 'users/dev' forums
[see below why I call them 'forums']. This means that development is
mostly done internally, or privately. Most available forums are
equivalent to our 'users' forums where power users post questions on use
and simply don't care about the internal development. I believe this
sums up the problem of acquiring new developers: people are rarely
sucked in and they don't get to 'know' and appreciate the developers by
listening to their dev-oriented conversations.
Star stages are bad, but visibility is the way people pay back. having a
mozilla.org account is not seen as valuable as having an apache.org
account. why? well, because all netscape people got one for free! there
is no clear meritocracy and this means no clear visibility value since
there are no mozilla stars. not even as a group. nothing.
This is emerging now with the smaller projects like phoenix and camino
and minotaur where a few people really drive the show and earn their
merit and create envy in others that want to be part of that show and
feel cool.
ego is an incredible motivator for geeks. only recently mozilla.org has
understood a way to make good use of it and this is stop forcing
everybody in the same room and see what happens.
[even the ASF is doing this by promoting projects in top-level domains
and this is a good thing for both, IMO]
- o -
Now, there is one difference that puzzled me.
In mozilla.org, forums are newsgroups. In apache.org, forums are mailing
lists.
Yes, I'm aware that it's possible and quite doable to automatically map
one into the other (and sites like GMane do already for some of our
lists), but I think the differences might be a lot more important.
Worth discussing the differences:
1) the ASF has human spam filtering, mozilla.org doesn't. This shows.
The amount of spam in their newsgroups is, well, irritating even if not
very high. This is clearly a plus for our approach.
2) mozilla.org is archived at google groups. this is clearly a plus for
their approach since our mail archiving and indexing capabilities, well,
suck ass compared to theirs. (no offense for the eyebrowse people, just
stating reality) moreover, google groups archives the entire history of
the internet. From an historical perspective this is going to be
incredibly important (and another reason to worry about google to be the
next microsoft/AOL big-brother-wannabe, but that's another story)
3) mail clients have newsgroups-advanced features that are normally
lacking in mail folders. For example, autotrimming forums to, say, the
last 500 messages. (please don't tell me how smart is your client or
your solution to do it anyway: my point is that newsgroups and NNTP were
*designed* for forums, while email was designed for point2point
communication and clients reflect these mindsets) also the need for mail
filtering is a lot decreased (again, don't tell me how you do it because
I don't care)
4) for newsgroups, lurkers can read archives from their favorite
mail/news reader. for mail lists, they have to use web-based intefaces.
This is, IMO, a *HUGE* plus for newsgroups.
- o -
Now, I'm asking: what if the ASF provides its own news server that wraps
around all our current mail lists setup and make them available to all
news-archiving services and news-reading clients?
Of course, it should be transparent and allow to redirect back a
newsgroup post to the mail list, which will then be handled by the ezmlm
manager, like done today.
So, you get the best of both worlds.
I really think this is the only thing we can learn from the mozilla.org
community so far. But it might be a good thing.
What do you think?
--
Stefano.
---------------------------------------------------------------------
To unsubscribe, e-mail: [EMAIL PROTECTED]
For additional commands, e-mail: [EMAIL PROTECTED]
