Hacker News Reader

Comment by daeken | Hacker News Reader

Ok, I have a lot of respect for esr, but all this talk of "porting" makes me think that no one knows how any of this works. "Porting" Android or WP7 to a new phone consists of: write drivers, write userland apps that give your users the experience you want, and possibly modify your existing bootloader code. They have documentation on their hardware, they have existing BSPs to reference, etc. This is not a hard problem. Getting this up and running with basic functionality would absolutely take no more than a month. Period.

View on HN

And then there's this bit:

>History matters; the Android codebase is designed to be ported in ways Microsoft is probably culturally incapable of even imagining.

NT is portable. It's been ported to at least four CPUs that I know of (x86, Alpha, MIPS, and PowerPC), and MS has announced that they will deliver it on ARM.

I don't much care for MS, but pretending that they're incapable of writing a portable OS is nonsense.

Um...you do know that all the non-Alpha/x86 ports were considered failures in-house and dropped, right?

Portability was one of the first things Microsoft sacrificed to get things "working right".

And the OS engineers who wrote the "portable" NT code have long since vested and moved on. The new kids on the block have probably never seen their code.

Right, and this is why my Windows NT 4.0 CD contained x86, Alpha, MIPS and PPC versions. I had never tried anything but Intel, but it was certainly not just in-house.

Right, and how many Windows NT apps were there for MIPS and PPC? If Developers still have too much trouble creating "universal" apps, your port is effectively dead. Outside of x86 and Alpha, NT was never used.

> Um...you do know that all the non-Alpha/x86 ports were considered failures in-house and dropped, right?

This is patently false.

> And the OS engineers who wrote the "portable" NT code have long since vested and moved on. The new kids on the block have probably never seen their code.

As an ex-Microsoftie I can tell you that you are wrong. Check Dave Cutler, Larry Wang and several other engineers (even from the original VMS team) still at Microsoft. NT has been ported to several different architectures internally many times over, though commercially only the x86 and PowerPC (XBox 360) variants have been widely released. Itanium ports worked perfectly well, but there was never any significant market for it.

They have it working on ARM internally already. And the current WP7 OS kernel is Windows CE, one of the most widely ported OS of all time, running on SH4, MIPS, x86, ARM i.e. almost any little-endian architecture.

Nokia's CEO shouldn't be focused on tech. He should be focused on significant internal cultural factors he was clueless about. He should also try to figure out how he and his advisers could've got it so severely wrong. After that, he has about 3 more Whys to go.

http://en.wikipedia.org/wiki/5_Whys

This has nothing to do with the CEO, it's responding to everyone talking about how long it's going to take to "port" WP7.

Nokia's current situation has everything to do with the C-level team.

Bringing up new hardware is a bit harder than that -- take a look at the most open example we have of doing so, the OLPC. 100% open hardware which they designed, but building the tools and bringing it up were still a lot of work.

Credit where it's due: the Android folks have worked very hard to make this an easy process for their OEMs.

By the way, this is why just about all hardware these days is brought up on Linux first. Yes, even your dear iDevices. :-)

I would have thought iDevices used BSD, since the whole Mach-kernel architecture is BSD based and all iDevices claim to run a Mach-kernel derived OS.

(I am aware there was MkLinux, which was also mach kernel powered, but it had very bloated performance compared to linux.)

It was my understanding that with phones, the bulk of the time is spent in testing the radio hardware to bring it up to compliance. That's why there is such a large delay between reference implementations and actual hardware. Software-wise, you're right, porting Android is only slightly less trivial than porting WP7

I think you're right. Look at the gap between WP7 GSM and CDMA launch. It's definitely not a trivial problem.

I think as developers unfamiliar with the tech we'd like to think that different radio stacks can be abstracted away, but the reality is likely very different.

you have a lot of respect for esr, I don't.

Eric wouldn't be able to port Android on his own to a platform like x86, much less a platform with a constrained environment (like a phone).

He's just not good enough.

So when he writes about the efforts involved, he doesn't know what the f---- he's talking about.

I was wondering who would be the first in this thread to engage in off-topic personal attacks against the author. You were the first. Congratulations, I guess?

Please, let's try to keep HN above this.

Reply on news.ycombinator.com