Android source available [LWN.net]

Google has finally released the source to its "Android" mobile phone platform; it can be obtained from source.android.com. It's not for the faint of heart: "The source is approximentely [sic] 2.1GB in size. You will need 6GB free to complete the build."

to post comments

Android source available

Posted Oct 21, 2008 18:14 UTC (Tue) by ikm (guest, #493) [Link] (18 responses)

"To build the Android source under Linux, you will need Ubuntu."

Nice!

I presume it's just what the developers are using...

Posted Oct 21, 2008 18:16 UTC (Tue) by khim (subscriber, #9252) [Link]

I don't think testing with bunch of distributions was their goal (that's what the community is good for) and 6.06 is LTS, so... makes sense.

Android source available

Posted Oct 21, 2008 18:27 UTC (Tue) by kragil (guest, #34373) [Link] (11 responses)

Ubuntu is what they and most people use. And it works with every Ubuntu release so it likely to run on Debian and most other Debian based systems. That covers a lot of ground.
Why should Google care for Red Hat or Novell? Or even Gentoo, Arch, Slackware or any other of the 500 distros?

Android source available

Posted Oct 21, 2008 19:22 UTC (Tue) by bcbarnes (guest, #51878) [Link] (10 responses)

And in the January 2018, as the distro wars and Gnome/KDE wars continue, people will continue to
say..."Really, this is the year Linux is going to make inroads into home desktops and corporate
networks."

I mean come on. If Google, a zillionaire open source supporting company, can only release their
premier platform OS for one linux distro, what hope is there for widespread adoption and
development?

Our entire lab uses Fedora and CentOS. I'd wager most of the HPC community uses a RHEL based
distro (if they use Linux). I guess the Comp Sci people prefer Debian/Ubuntu. What's the point of
open source if a company like Google can't even make cross-linux-platform releases?

Android source available

Posted Oct 21, 2008 19:24 UTC (Tue) by bcbarnes (guest, #51878) [Link] (1 responses)

Well that was weird. I'm posting via Safari right now and none of those linebreaks were visible while
composing or previewing. Here, I'll post a long paragraph again. No visible linebreaks while
previewing the comment.

Android source available

Posted Oct 21, 2008 20:02 UTC (Tue) by nix (subscriber, #2304) [Link]

'Plain text' automatically linebreaks for you.

That's weird

Posted Oct 21, 2008 19:34 UTC (Tue) by khim (subscriber, #9252) [Link] (2 responses)

When IBM (company with much longer involvemen in Linux) makes Fedora 7-only SDK there are no problems, but when Google supports single distribution it's weird??? Come on, it's not that unusual: it's pretty common for platform kits to be pretty rigid. You need Windows 2000 SP4 only (don't try Windows XP or Vista) and package must be installed on D:\ or "bad thing will happen" - that's typical instruction I've seen.

as soon as someone contributes patches ...

Posted Oct 21, 2008 20:56 UTC (Tue) by JoeBuck (subscriber, #2330) [Link]

... I'm sure it will then build just fine on other distros.

That's weird

Posted Oct 21, 2008 23:40 UTC (Tue) by robert_s (subscriber, #42402) [Link]

"When IBM (company with much longer involvemen in Linux) makes Fedora 7-only SDK there are no problems,"

Who said there were no problems?

Android source available

Posted Oct 21, 2008 20:40 UTC (Tue) by floop (guest, #5889) [Link]

I don't understand why a person would care about the release of the code in general, besides "hey that's great for oss", if they weren't capable of getting it to work under Fedora or Gentoo if they wanted.

Android source available

Posted Oct 21, 2008 20:49 UTC (Tue) by thoffman (guest, #3063) [Link] (1 responses)

Dude.

This is hardly a consumer product. We're talking about a developer SDK here. Now, if Google released some consumer level Linux software, (e.g. Picasa) that only worked on one distro, I could see a little more cause for complaining, but this is for developers.

If a developer can't either:
(1) set up a VM or dual boot or new machine with Ubuntu to work with Android, or,
(2) Cooperate with the community to fix whatever little issues are required to make the Android SDK work on their distro of choice

then... how useful a developer are they anyway?

Android source available

Posted Oct 22, 2008 15:27 UTC (Wed) by davidw (guest, #947) [Link]

It's not the "SDK" - that is available, like it should be, for Mac, Windows and Linux. The SDK is what you need to write programs for Android. This is the source code release - what you need if you want to rebuild the whole deal. Not many people really need that, and hey - it's open source isn't it, I'm sure some enterprising Fedora users can hack it to work there and send in patches.

Android source available

Posted Oct 22, 2008 15:24 UTC (Wed) by sbergman27 (guest, #10767) [Link] (1 responses)

Are the critics right then? Has Linux so fragmented (like Unix did) that "Linux" users must have these arguments over which distros should be supported? Or is this just another excuse to attack Ubuntu. I use mostly CentOS on my servers, and Ubuntu on my home desktop. They are both exellent distros, but they specialize in different areas. Why do critics of Ubuntu operate on such a hair trigger?

BTW, is your lab, and a significant portion of the HPC community planning on building mobile phone OSes? On existing equipment? If you, and they were going to, would you, and they be capable of installing Ubuntu?

Or are Linux installation procedures so fragmented, as well, that expertise in installing CentOS does not transfer to installing other Linux distros?

Android source available

Posted Oct 23, 2008 2:15 UTC (Thu) by drag (guest, #31333) [Link]

> Are the critics right then? Has Linux so fragmented (like Unix did) that "Linux" users must have these arguments over which distros should be supported? Or is this just another excuse to attack Ubuntu.

Both.

----------------------

It used to be that Linux distributions differed quite a bit for very significant reasons. Some provided configuration tools and interfaces that were proprietary in a attempt to do the 'value added' approach to getting licensing costs from linux. Some people dissagreed about how you should approach package management systems and so on and so forth.

Nowadays everybody uses pretty much the same stuff. The major differences are going to be Redhat configuration tools vs Debian's package management system. Other then that people are switching to upstart, they are using network-manager for desktops, dbus, packagekit, policykit. Pretty soon they should be using about the same initramfs environments and so and so forth. Much more the same then different.

IMO one of the major goals for distributions is to eliminate the differences in the so-called 'Linux plumbing' and end up using, more or less, identical systems on the low-level. Then use the same core as a basis to then branch out and do their own thing. Update packages, experiment, etc etc.. but always each time they do a release they re-base off the same core system that they share.

This, I think, will end up going to make application developer's (open source and otherwise) lives a lot easier, as well as system integraters and people that need to document how the system works for normal folks.

....

As far as Ubuntu-hate goes. It's pointless and misdirected.

Not limited to Ubuntu

Posted Oct 21, 2008 22:49 UTC (Tue) by BrucePerens (guest, #2510) [Link]

You are welcome to make it build under something else, and send back patches.

Android source available

Posted Oct 22, 2008 10:09 UTC (Wed) by kripkenstein (guest, #43281) [Link]

I think it's perfectly reasonable to not support all distros, *if*

1. At least one significant distro is supported completely, i.e., it works there (this is then a base to start from), and

2. Google cooperates with other distros in getting it to work with them as well, if/when those distros are serious about doing so, and

3. No code/build scripts/etc. are made in a way that would make them hard to run on other distros.

We already have (1), and I hope that (2) and (3) as well. If so, then I find no fault here.

Now, if this *wasn't* open source, then I would have very different criteria. In particular, I would expect at least the major distros to be supported by the vendor. (Of course I prefer if it's open source, but I'm just contrasting. For open source, I expect less from the vendor because we the community can and should do part of the work.)

Android source available

Posted Oct 23, 2008 5:32 UTC (Thu) by yanfali (subscriber, #2949) [Link] (1 responses)

I actually built the code on gentoo just fine...

Failing to build on $RECENT_DISTRO would be a bug

Posted Oct 23, 2008 10:39 UTC (Thu) by alex (subscriber, #1355) [Link]

Seeing as this is an embedded product I would be surprised if it didn't build on all but the most broken Linux distro. I assume the only real dependencies will be having an appropriate cross-compiler.

If there are any dependencies on the build host's installed libraries I would expect that to be a bug.

Android source available

Posted Nov 5, 2008 12:21 UTC (Wed) by Amit009 (guest, #55016) [Link]

I dont think so we can make in any platform

Involvement

Posted Oct 21, 2008 18:19 UTC (Tue) by sladen (guest, #27402) [Link] (4 responses)

People getting involved, comes over they get the thrill that they can tweak and improve.

If somebody can build-test cycle the whole thing on their laptop in under five minutes, they tend to be more likely to be involved and more likely to join the club.

Fast build-test-debug means small chucks, which means modularity; cf.:

a fairly large and (by default) monolithic office productivity suite
a suite of modularised graphics card display drivers
a retargetable operating system kernel where you can turn most of the code (drivers) off, or just stick to your plugin

I wonder which this 2 GB tarball is going to turn out to be?... And as a result, how many [willing] contributors it is going to have.

Involvement

Posted Oct 21, 2008 18:29 UTC (Tue) by marduk (guest, #3831) [Link]

Agreed. This sounds less of a "Here it is. Dig in!" and more of a "Here is is. Now we can say it's OSS!"

You have pretty powerful laptop I'd say

Posted Oct 21, 2008 18:39 UTC (Tue) by khim (subscriber, #9252) [Link] (2 responses)

If somebody can build-test cycle the whole thing on their laptop in under five minutes, they tend to be more likely to be involved and more likely to join the club.

Can you do this with Firefox? Or with kernel? Or with GNOME (let's forget about KDE)?

This thing looks pretty modular (they are using multiple Git's repositories!), supplied scripts only make it easier to compile and build the whole thing without hassle.

That being said it's not just a bunch of well-known open-source libraries. That is kinda the point...

You have pretty powerful laptop I'd say

Posted Oct 21, 2008 18:54 UTC (Tue) by tzafrir (subscriber, #11501) [Link] (1 responses)

GNOME and KDE are indeed much more modular. They are composed of many sub packages, and you can have a partial build of those separately (and fix separately.

The kernel tree is huge, but if you just want to quick-fix a module rather than the core, you can do that. Still, this "big" amounts to 200MB of disk space required (or so).

I'm not sure about Firefox,

You have pretty powerful laptop I'd say

Posted Oct 21, 2008 19:14 UTC (Tue) by ibukanov (subscriber, #3942) [Link]

> I'm not sure about Firefox,

For Firefox you need 700 MB for sources and 1.2GB for the build. This is on Fedora-9.

A full rebuild of the browser takes 11 minutes on 4-core 2.6 GHz Xenon or about 80 minutes on 1.1 GHz single-core laptop. But after that minor patching typically leads to 3-5 minutes of wait-time even on the laptop or some seconds on that server using official build instructions. With some knowledge of the build system and depending on the subsystem, one may cut that incremental build time by factor of 10.

Android source available

Posted Oct 21, 2008 19:10 UTC (Tue) by lacostej (guest, #2760) [Link] (3 responses)

ant the code is hosted on android.kernel.org. Cool.

Android source available

Posted Oct 21, 2008 20:43 UTC (Tue) by armijn (subscriber, #3653) [Link] (2 responses)

Odd. I thought kernel.org was supposed to be vendor neutral. I don't think Android is vendor neutral, it is just Google's phone platform.

I think that's EXACTLY why

Posted Oct 21, 2008 21:29 UTC (Tue) by khim (subscriber, #9252) [Link] (1 responses)

Odd. I thought kernel.org was supposed to be vendor neutral. I don't think Android is vendor neutral, it is just Google's phone platform.

Actually no, that's OHA's platform. I think that's the reason. I'm pretty sure http://code.google.com has enough horsepower to handle the load, but other participants wanted some place not controlled by Google. At the moment a lot code out there is Google's creation, true, but the idea is that this situation is temporary.

Let's see how well it'll go. Results of previous such tries (RedHat with Fedora and Sun with OpenSolaris) are certainly mixed...

I think that's EXACTLY why

Posted Oct 22, 2008 9:59 UTC (Wed) by kripkenstein (guest, #43281) [Link]

> I'm pretty sure http://code.google.com has enough horsepower to handle the load, but other participants wanted some place not controlled by Google.

I think the reason might be that Google Code can't handle git, only subversion (last I heard. If that changed, then ignore this comment).

Android source available

Posted Oct 21, 2008 19:34 UTC (Tue) by rfunk (subscriber, #4054) [Link] (10 responses)

Now if only an Android-compatible phone were available from a US carrier
with more coverage than T-mobile......

Android source available

Posted Oct 21, 2008 20:39 UTC (Tue) by kragil (guest, #34373) [Link]

Patience, young padawan.

T-Mobile

Posted Oct 21, 2008 21:42 UTC (Tue) by linuxrocks123 (subscriber, #34648) [Link] (8 responses)

Not sure where you live, but I have T-Mobile and coverage is just fine. Now, service is pretty bad (dropped calls are very common), but I'm usually able to connect to a T-Mobile base station with a strong signal. And, when I can't, my phone is usually able to associate with another carrier with whom T-Mobile codeshares.

T-Mobile

Posted Oct 21, 2008 22:06 UTC (Tue) by rfunk (subscriber, #4054) [Link] (7 responses)

Sure, their coverage is fine in a lot of places, mostly in cities. I'm 20 miles outside the city, though right on an interstate.

I checked T-Mobile's coverage map, and their coverage all around where I live (except a nearby town) is at the minimal "1-bar" level. AT&T's coverage is at their "best" level all around here, and Verizon (which I have now) claims the same for an even wider area. Even Sprint, not known for great coverage outside urban areas, claims much better coverage out here than T-Mobile does (though there does seem to be a hole right around my house).

I assume that AT&T doesn't feel the need for a really nice Android phone when they have the iPhone; meanwhile, Verizon wants to exert way too much control, the opposite of the Android ethos. (Plus their wide coverage keeps people coming back all by itself, so they don't feel the need for as much of a competitive edge with their phones.) I imagine Sprint will have an Android phone at some point, but I doubt the top two will go for it.

T-Mobile

Posted Oct 22, 2008 7:12 UTC (Wed) by drag (guest, #31333) [Link] (5 responses)

AT&T and Tmobile are using GSM networks.

You can use whatever phone you want with those. All you need is to find a GSM phone that can run Android and your set. I don't know if any are out yet, I doubt it, but once they are available then you can go out and do whatever you want.

With GSM your service is bonded with the GSM chip your given. You can purchase a Go-phone, for example, with a pay-as-you-go service contract, pull out the GSM chip, and use that same service with a different phone.

With CDMA networks it's much more difficult since you'd have to register your phone and configure it to work with a specific vendor. You can't just mix and match whatever you want.

The same would be true for the IPhone (being GSM) if Apple were not complete nazis about it and had the DRM'd up the you-know-what. (of course there are hacks, but whatever)

The thing is, of course, the T-Mobile stuff is subsidized by your service contract. So the 'purchase price' of the phone is really the 'down payment' and you pay installments every month for your phone along with your regular phone bill. So it seems the phone is much cheaper then it really is. (for example: the IPhone's price tag is a simple out and out deception)

Now if you pay full retail price for your phone then you can use whatever service contract you want with any GSM provider. There are special cheaper service contracts and bundles that AT&T and friends won't advertise on their websites or tell you about in their stores. They exist, but you have to poke and prod to find out about them.

GSM

Posted Oct 22, 2008 14:24 UTC (Wed) by rfunk (subscriber, #4054) [Link] (3 responses)

Yes I know all that (except the "special cheaper service contracts" part),
but so far the only Android phone (a highly capable one) is available only
with a T-Mobile contract.

GSM

Posted Oct 22, 2008 14:36 UTC (Wed) by drag (guest, #31333) [Link]

Ya. I just like to point that stuff out because people just have the assumption there is a inherent 'lock-in' to US cellular networks that don't quite apply to all carriers and phones.

Personally this is the sort of thing I've been waiting for:
http://www.appleinsider.com/articles/08/10/20/intel_shows...

Full screen devices. The controls integrate into the display so it makes it possible for all sorts of stuff, especially games, which is something that is not very good on today's Cell phones compared to special-purpose devices like the PSP or Nintendo DS.

With Intel's commitment to Linux lets hope that those things will be open enough to run generic Linux systems as well as Android.

Actually T-Mobile's CTO said you can just buy it...

Posted Oct 22, 2008 21:08 UTC (Wed) by khim (subscriber, #9252) [Link] (1 responses)

Here are details. You can just buy it for $399 and go with any carrier.

Don't know how well it'll work on practice...

Actually T-Mobile's CTO said you can just buy it...

Posted Oct 22, 2008 21:32 UTC (Wed) by rfunk (subscriber, #4054) [Link]

Good to know, thanks for the useful link!

T-Mobile

Posted Oct 30, 2008 15:31 UTC (Thu) by kamil (guest, #3802) [Link]

> Now if you pay full retail price for your phone then you can use whatever service contract you want with any GSM provider. There are special cheaper service contracts and bundles that AT&T and friends won't advertise on their websites or tell you about in their stores. They exist, but you have to poke and prod to find out about them.

Sorry for going off-topic, but can you provide more info, or links, about that?

T-Mobile

Posted Oct 22, 2008 8:07 UTC (Wed) by Guhvanoh (subscriber, #4449) [Link]

This just in from the rumour mill. See url:
http://www.engadget.com/2008/10/20/motorolas-android-slid... http://www.engadget.com/tag/OHA has more links on the this. As for T-Mobile, when I lived in Baltimore, MD their coverage was there but the dropped calls made me drop them. In London, UK I had the same problem. So I dropped them. They were cheaper, but I wasn't paying for dropped calls.

Android source available

Posted Oct 21, 2008 22:13 UTC (Tue) by leoc (guest, #39773) [Link]

So is it actually possible to install this code on a phone?

Android source available

Posted Oct 22, 2008 6:01 UTC (Wed) by allesfresser (guest, #216) [Link] (1 responses)

I followed the instructions to grab the code (using Ubuntu Hardy x86-64), and I got a bit of code, and then this result:

Initializing project platform/dalvik ...
warning: no common commits
remote: Counting objects: 6557, done.
remote: Compressing objects: 100% (3524/3524), done.
remote: Total 6557 (delta 2423), reused 6557 (delta 2423)
Receiving objects: 100% (6557/6557), 6.81 MiB | 907 KiB/s, done.
Resolving deltas: 100% (2423/2423), done.
From git://android.git.kernel.org/platform/dalvik
 * [new branch]      master     -> korg/master
 * [new branch]      release-1.0 -> korg/release-1.0
fatal: The remote end hung up unexpectedly
error: Cannot fetch platform/dalvik
 /arc/mydroid $ repo sync
fatal: The remote end hung up unexpectedly
error: Cannot fetch platform/build

Has anyone else seen this? I don't have much (OK, any) experience with git, so maybe this is not uncommon, but I don't know.

Android source available

Posted Oct 23, 2008 5:34 UTC (Thu) by yanfali (subscriber, #2949) [Link]

seems like their git repositories are just busy. Just keep retrying. I had to restart repo a bunch of times till I got a connection.

Droid fonts

Posted Oct 22, 2008 7:53 UTC (Wed) by nim-nim (subscriber, #34454) [Link] (1 responses)

Has the licensing of the droid fonts finally been clarified or are they still a binary blob hidden deep in a monster archive with no clear license attached?

Droid fonts

Posted Oct 22, 2008 16:36 UTC (Wed) by yosch (guest, #4675) [Link]

Sadly it seems the situation hasn't evolved much :-(

The fonts are in the base.git repo under data/fonts. There are README and Notice files mentioning the Apache v2 license but this is not reflected in the font metadata. The font metadata actually contains contradictory information!

The Ahem test font does not have any license information attached to it.
No way of knowing who the upstream is. A FONTLOG should really be added.

And the Droid fonts themselves still refer to an unknown external EULA in the License Description field:

"License Description: This font software is the valuable property of Ascender Corporation and/or its suppliers and its use by you is covered under the terms of a license agreement. This font software is licensed to you by Ascender Corporation for your personal or business use on up to five personal computers. You may not use this font software on more than five personal computers unless you have obtained a license from Ascender to do so. Except as specifically permitted by the license, you may not copy this font software.If you have any questions, please review the license agreement you received with this font software, and/or contact Ascender Corporation. Contact Information:Ascender CorporationWeb http://www.ascendercorp.com/"

A standard link to the Ascender restricted EULA is provided in the License Info URL field: http://ascendercorp.com/eula10.html

You can update these fields using Fontforge:
http://fontforge.sourceforge.net/fontinfo.html#TTF-Names

Not the kind of choice that will build a collaborative font designer community around this to make the Android stack more multilingual...

I really doubt many font designers have heard about or read through the Apache v2 license. IMHO it's in the interest of Google and the Android community to pick a community-recognized font-specific license like the OFL to provide a better collaboration mechanism to extend the font and to take care of embedding/naming problems.

We need to get in touch with the Android folks again...

Android source available

Posted Oct 22, 2008 8:02 UTC (Wed) by johill (subscriber, #25196) [Link] (4 responses)

There's little, if no, value in this code. Here's a select snippet:

        fd = open("/dev/mem", O_RDWR | O_SYNC);
        if( fd == -1 ) {
            print_error("Cannot access /dev/mem\n");
            goto init_driver_end;
        }
        nvsPtr = (volatile unsigned long *)mmap(0,0x1000,PROT_READ|PROT_WRITE,MAP_SHARED,fd,0x13F13000);
        eeprom_image_length = *(nvsPtr + (0xE0C >> 2));
        print_debug("---Eeprom from Memory Size = %u\n", eeprom_image_length);

or how about kernel code that is licensed under "Apache License, Version 2.0" which is clearly incompatible with the GPL2...

Android source available

Posted Oct 22, 2008 8:15 UTC (Wed) by johill (subscriber, #25196) [Link]

I stand corrected, the apache 2 license code I found was some userspace tool. Not that the kernel code is any decent quality so that it would be suitable for merging without complete rewriting.

Fastest review ever

Posted Oct 22, 2008 10:43 UTC (Wed) by avik (guest, #704) [Link]

"There's little, if no, value in this code"

This has to be the fastest review I've ever seen of a 2.1GB code base. I'm sure you wouldn't like your own code to be dismissed out of hand like this; please show more respect towards others'.

Android source available

Posted Oct 22, 2008 13:18 UTC (Wed) by busterb (subscriber, #560) [Link]

This code is useful in the following ways:

1. Now we know where the eeprom is in the memory map of a G1 phone, its size and as well at
least one of its elements.

2. I've know a few embedded developers who try to write an IOCTL-driven character device to
access the most trivial one-off hardware. Here, we can see the Android developers are at least
pragmatic about embedded development. Maybe some embedded developer will finally slap his
head and say 'of course - why was I going to all that trouble with /dev/eeprom when /dev/mem
was just as useful'

3. This code obviously has to run as root - somebody can audit the code now for security
problems.

Android source available

Posted Oct 23, 2008 12:58 UTC (Thu) by ikm (guest, #493) [Link]

> init_driver_end;

I presume that code is a part of some userspace driver. As such, what is exactly wrong working directly with /dev/mem? If I remember correctly, the approach of userspace drivers is an officially blessed practice and is actually recommended for embedded developers.