Stone Mirror wrote:

I interpreted Mr. Morril's statements to mean that Google does not expect other existing open source projects, such as GNOME Mobile, to drop what they're doing and switch to Android en masse. Android, once released as open source, may well develop its own community of interested folk, just like any other open source project of significance. GNOME itself, for example, probably didn't expect the KDE folk to abandon their efforts and run to GNOME, yet GNOME has done quite well for itself.

New initiatives in existing markets are par for the course in open source. For example, in your two messages on this thread, you have cited several open source mobile projects and didn't even get them all (e.g., Maemo). By your argument, most of those shouldn't exist, since they're duplicating efforts of other such projects.

Similarly, new open source programming languages should not exist because they are "in opposition to the existing, mainstream (language) open source community", which might irritate the Groovy and Scala folk, to name two.

Heck, on the front page of Sourceforge.net right now, three of the top six "most active" projects are ERP implementations, presumably with some amount of overlap.

Some newcomers to a space succeed nicely (e.g., GNOME busting into KDE's space). Some newcomers to a space don't fare quite so well (e.g., Helix busting into mplayer's space). But it does happen, and with some frequency. Android, in all likelihood, won't be the last entrant into the mobile open source space; they're just the next.

That depends on the code. Utility JARs (e.g., JavaMail) that use Android's subset of Java SE work just fine.

Existing Swing or SWT code will not. But, by the same token, when SWT was released, existing Swing code wouldn't work on it, and when JavaME code was released, existing Swing code wouldn't work on it, and when ZK was released, existing Swing code wouldn't work on it, and so on. Life, amazingly enough, went on, and those technologies each received some measure of adoption (though, in some cases, not as much as they might have liked).

First, "fewer than 1,800 applications" apparently only counts ADC entries. Not all applications built for Android were submitted to the ADC. In fact, for a platform with zero devices in the wild, ~1,800 applications is pretty damn impressive. AT&T would have sacrificed many a goat to have had ~1,800 HDML-capable Web sites when their first mobile Web phones came out in the mid-1990's.

Second, you assume that the reason the number of SDK downloads dwarfs the number of applications submitted to the ADC is because of the learning curve. There are undoubtedly many other contributors to that disparity, including:

-- The fact that there are no devices and therefore no market (commercial or open source) for applications at this time

-- People deciding that Android isn't their cup of tea for reasons other than learning curve (e.g., didn't realize it was just Java, don't like the toolset)

-- People downloading the SDK just because it's from Google, before realizing they really didn't have anything much they wanted to do with it, anyway

There is a learning curve to Android, to be certain, but it's not significantly worse than many other Java frameworks I've dealt with.

Well, you *did* choose to link to El Reg, which isn't exactly a bastion of unbiased reporting... ;-)