Harmony…

The thing about Apache Harmony. Well, there are a lot of problems with Apache Harmony, but the thing about Harmony that is going to be a real problem for them is that I don’t see any way for their development process to sustain itself. Which development process largely consists of batting their eyes and posting gibberish to mailing lists, and occasionally getting someone to donate chunks of code.

To that point, they should have called the project “Apache Panhandle,” because they aren’t interested in any kind of Harmony with any other free Java hackers.

And so Intel dropped a bunch of AWT code on them recently. This is a huge step forward for them, and the indication is that they’re at more that 80% of 1.4 compatibility. That’s really nice, and it’s a huge leap forward for them. But that excitement doesn’t cover exactly how they plan to finish that last 20% (and more like 30% for 1.5), especially since that last mile is really going to be the hardest one to cover. It’s always like that, and anyone reading this knows it, such that it’s kind of dumb and obvious to point it out — the last 10%, the last 1%, are going to be the hardest to finish and will take the most time. And if Harmony’s plan is to continue to wait for code donations to magically appear, they’ll happily remain at 80% until we all finally forget about them.

Roman also pointed out another thing that’s weird and unsustainable about Harmony’s development: that it mostly seems to happen in secret, with occasional giant code drops. How is that free or open? And more importantly, how the hell can you coordinate that? Managing disparate groups is hard enough; doing when no-one does anything in the open is even harder.

I’m done ranting for now :-).