Android is ultimately intended for OEMs. If Google was not dealing with corporations that have billions of dollars at stake, they might be more likely to be more open. But honestly, the difference is the product. A browser is free to the final customer, has no real hoops to jump through before release, and since all browser makers have been heading to the same basic goal for over a decade and (mostly) adhere to a standard, there is not really anything to gain from secrecy. A phone/tablet costs money to the final user, has to go through OEMs, carriers, and the Feds before it can be in anyone's hands. Not to mention that we are in the infancy of mobile computing and secrets can give a real advantage. I wouldn't be surprised if they are much more open in five years.
Basically, I'd like them to be more public, but there are real reasons not to be currently and a code dump is better than nothing.
The Chrome team doesn't fully work in public either. For example, Crankshaft was developed in secret and then dumped into the bleeding-edge branch on the day it was announced.
For each Android release one phone vendor seems to get an exclusive; presumably they are paying for that in some way (like TV ads that promote the platform).
But since you can't see the current state of development, you can't know whether your fix is even relevant anymore. That's a big reason why I prefer to use and contribute to CyanogenMod, and if Google ever wants to cherry-pick my patches (minor though they are), they're welcome to it.
No the openness argument is that the code is open and it is. It is 100% open source, Apache and GPL. What this does is limit their openness. You can't see every line as it's written.
I would guess they do it this way because they work with a lot of unreleased hardware, and the carriers and manufacturers don't want any leaks.
Lots of programmers prefer to tinker in private and only submit to repositories after everything is cleaned up and debugged. Others are fine with exposing every keystroke. In both cases we should be grateful if the end result contributed to the public good as Free Software.
Wait, I'm not sure I understand: "the openness argument is that the code is open and it is" and "this does is limit their openness" -- what do you mean by that?
Anyway, my point is that you would not call Firefox an open source project if it was developed internally by Mozilla and just released code every major release. It's not that I don't think it's good that that they release the code, it's that they use "open" as an argument when it's only open when it benefits them -- and not a second before.
I doubt that this shouldn't have been here on HN -- Google knows and anticipates that all news around Android are being watched closely and heavily anticipated by a large and tech savvy audience. I'm just not sure what to think about the current release delay. I have a hard time believing Google just pushes the code to kernel.org without contributing network infrastructure and servers... couldn't they easily help out kernel.org if this was just a technical problem.
Isn't this what bittorrent is for? Or any number of other peer-to-peer services. I mean, maybe everyone will need to verify with a MD5 or SHA. But this seems like such a 90s way of distribution. It's too bad pirating has given peer-to-peer such a bad name. Ideal valid use cases for P2P, like this, end up avoiding it due simply to social and cultural influences.
It would have been wiser to post a message saying "I'm done committing Gingerbread." It was inevitable that announcing it beforehand would cause enough traffic to bring the server crashing down.
