In my previous post I covered what a toolchain is and why you need one to cross-compile applications for the iPhone.

I’ve seen rumors that there will be a Windows toolchain soon, and in theory you can create an iPhone toolchain on older Apple computer with PowerPC chips, but most of the iPhone development these days seems to favor the newer Intel-based Apple computers.

One of the best resources for all iPhone related info is the iPhone dev wiki at http://iphone.fiveforty.net/wiki/index.php/Main_Page (unlinked because the wiki maintainers aren’t sure that they can handle lots of visitors and request that people not link directly). The toolchain page at http://iphone.fiveforty.net/wiki/index.php?title=Toolchain_Project has a lot of details, and the talk page (look for the link labeled “discussion”) gives even more background. But those pages are really terse.

Here’s a little more background info, and I’ll include some additional links at the end.

Getting gcc

Did you know that recent Apple computers (and even the iPhone) ships with a solid UNIX-based kernel? You might have heard that. Something I didn’t know is that Apple computers don’t ship with a compiler such as gcc pre-installed. It should be on the DVD that came with your computer; Apple calls its developer tools Xcode, and Xcode includes gcc.

You can also download Xcode/gcc from the web. Visit http://connect.apple.com/ and sign up for a (free) developer account. After you sign up, the “Downloads” link will become clickable. Click that, then click the “Developer Tools” link. I downloaded Xcode 2.4.1, although there is a beta of Xcode 2.5 available as of today. The download is a DMG file, which stands for “Disk Image.” Double-clicking on the .dmg file will mount the disk image and open a file folder that includes the file “XcodeTools.mpkg” and if you double-click on that package, you’ll get a window that guides you through the installation of Xcode Tools. When you’re done, you can type “which gcc” in a Terminal window and you’ll see that gcc is installed.

Downloading toolchain code

Are you tired yet? Then you might want to take a break, because things are just getting started. For example, the first thing you’ll read on the wiki toolchain page is:

To use the new toolchain, check out the latest branch (as of this writing, 0.20):

svn checkout svn://svn.berlios.de/iphone-binutils/branches/alpha/0.20/

And you might be thinking “What the heck is svn?” It’s a program called Subversion, and it lets you check out source code across the web. Here’s a page about how to install Subversion on a Mac. The short answer is that there’s a couple ways. First, you can install a program called Fink that in turn helps you install more UNIX-related programs such as Subversion. Or you can download a Disk Image file, click the .dmg file, and install svn directly.

Getting Libstreams

The next thing to do on the iPhone dev wiki page says

Download libstreams from Apple’s web site, compile it, and install it.

And I’m thinking “Could they get any more terse?” Plus the dev wiki is locked down to prevent wiki vandalism, so only a few people can edit that wiki. Thanks a lot, wiki spammers, you jerks. Now everybody has to interpret terse instructions on their own and can’t update the wiki with more detailed instructions.

To get/install Libstreams, it looks like you can fetch the correct source (PowerPC/PPC or Intel/x86) from
http://publicsource.apple.com/darwinsource/10.4.9.x86/Libstreams-24.1/ or
http://publicsource.apple.com/darwinsource/10.4.9.ppc/Libstreams-24.1/

You’ll have to log in with an Apple ID. Once you log in, you see about 20 files. Download each one in turn into a single directory, and make sure that you save the files as raw files, not html. Do that by mousing over a file, doing a control-click on the filename, select “Save Link As…” and make sure to change the “Format:” drop-down selector from “HyperText” to “All Files.”

Once you have all the libstreams files in one directory, open a Terminal window, cd into that directory, and type “make” to make the libstreams.a library. Then type “sudo make install” to install the library into the right location on your Apple computer.

Are you still with me? Because we’re really just getting started. Next the wiki says

Get a copy of the iPhone system software, and set the environment variable HEAVENLY to its location (export HEAVENLY=/path/to/iphone/software).

Unless you’re a real geek, you’re probably thinking “What the @#$% does that even mean?” Well, review my toolchain and cross-compiling post. In order to build a proper toolchain, we need some of the software that is only found on the iPhone (header files? libraries? I’m not 100% sure). Since we’re cross-compiling on (say) an Apple computer, that means that we need iPhone software on our Apple desktop machine. That’s a little bit of a problem. In theory, you could copy your iPhone’s filesystem to your computer. There’s even a program called Toolchain Helper mentioned at http://iphone.fiveforty.net/wiki/index.php/Toolchain_Helper that does that. But Toolchain Helper doesn’t run on a pristine iPhone; you need to “jailbreak” your iPhone first. You can also run AppTapp/Installer.app to let you install the Toolchain Helper.

There is another way to get the iPhone software onto your computer. Apple provides the disk image (DMG) file for iPhone software. Woohoo! Except that it’s encrypted. Bleah. But someone figured out how to decrypt the software! Woohoo! Except that some people worry that decrypting the software might be a violation of the Digital Millennium Copyright Act (DMCA). Bleah. But lots of people believe that reverse-engineering software for the purposes of interoperability is legal! Woohoo! And the DMCA contains an explicit exemption for unlocking cell phones in some circumstances. Woohoo! If you’re at all nervous about decrypting a DMG file, then don’t. Consult with your local lawyer to see what your comfort level is.

The best instructions I’ve seen to extract the iPhone DMG contents are Landon Fuller’s post. He has a link to some modified source code to a program called vfdecrypt, but for some reason when I compiled vfdecrypt it wouldn’t work for me on the DMG file. There’s a precompiled vfdecrypt file to be found at http://ellkro.jot.com/HowToBuildObjCApps inside of the phonedmg12.tar.gz file on that page.

And that, sadly, is as far as I’ve gotten so far. I haven’t managed to compile a working toolchain myself yet. I’m using an older PowerPC Mac, and trying to compile things caused errors for me. I enjoyed playing around with toolchains for a day, but I don’t have much more time to invest in this. It’s clear that building a toolchain is still not for the weak of heart. On the bright side, once you have a working toolchain, it looks like there are a ton of cool applications you could write.

All those draft blog posts were starting to bug me, so I decided to do some spring cleaning this weekend. A lot of the draft posts I just deleted. I transferred some stuff into personal Google Drive files. I was left with a dozen or so blog posts that mostly fit into the “very boring” and/or “half-finished” category. But I kept thinking that 1-2 people out of the two billion or so people online might actually run across a post and find it helpful.

So I’m going to just throw a few of semi-boring, semi-finished posts onto my blog. Feel free to ignore these.

I am currently an employee of Google. I receive a salary from them and I also own Google stock and options.

Other than compensation from Google, I don’t accept any money or other gifts of value from any companies or individuals. I don’t accept speaking fees, consulting fees, honoraria, or trips. I don’t accept free, discounted, or loaned products. When I receive unsolicited gifts of value from companies or individuals in the scope of work, I give away those gifts.

When I speak at a conference or event, I generally do not pay a registration fee for that event. Some conferences also waive registration fees for that event for one or more of my colleagues or a traveling companion. Either my company or I pay my own travel and hotel expenses when I speak at an event.

I do not run advertisements or otherwise receive any monetary compensation from the operation of my website.

Added January 16, 2010: A few years ago my wife and I formed a non-profit foundation. Neither of us are paid a salary from the foundation. Example groups that the foundation has donated to include the Electronic Frontier Foundation, MAPLight, Change Congress, the Sunlight Foundation, Free Press, the Poynter Institute for Media Studies, Committee to Protect Journalists, Public.Resource.Org, Khan Academy, Code for America, charity: water, and Room to Read. The Employer Identification Number (EIN) of our foundation is 203865461.

I have invested in Perfect Third (the company that makes the WakeMate), Zencoder, Cardpool, Tasty Labs, Drchrono, Grubwithus, PoundPay, Apportable, Mailgun, and Parse. I have also invested in Lowercase Capital (Lowercase Ventures Fund I), Y Combinator (Y Combinator Fund II), and Lowercase 140.

You don’t need to %SI3_rnd10 rod’s %SI3_rnd11 and %SI3_rnd12 %SI3_rnd13’ jokes!
This is a %SI3_rnd14 for %SI3_rnd15 your %SI3_rnd16! It will %SI3_rnd17 in seconds after she %SI3_rnd18 and %SI3_rnd19 as good as if it was a %SI3_rnd20 rod!
No more jokes – you will always get %SI3_rnd21 and moans! The huge pack costs less than 30 %SI3_rnd22!
%SI3_rnd23 can be a %SI3_rnd24! No one will know about your %SI3_rnd25!
%SI3_rnd26 now and save more than $10 regardless of your order’s size!

I think it’s spam about embiggening a specific body part. But the spammer clearly didn’t set up their spam template correctly. Anyone have guesses about which email spamming software package this is?

I plan to talk sometime soon about what I learned in the process of moving to a completely different domain for a month.

- I took one of my domains, dullest.com, and moved it to TigerTech from pair Networks.
- I installed the latest version of WordPress on dullest.com and copied the MySQL database from mattcutts.com to dullest.com.
- I changed my blog layout to the excellent Thesis theme by Chris Pearson. Previously I was using the Almost Spring theme.
- I added an .htaccess file that will do 302 redirects from www.mattcutts.com/path/file.html to www.dullest.com/path/file.html .

Note: changing your IP address, webhost, domain name, blog template, and blog version all at the same time is the exact opposite of what you should normally do. It’s better to change only one thing at a time so that if something goes horribly wrong, you can trace what caused it.

Also, if you were truly moving a site, a 302 redirect wouldn’t be the right redirect to use–a 301 (permanent) redirect would be better. But if I like these changes, I might migrate mattcutts.com to TigerTech and then migrate my blog from dullest.com back to mattcutts.com. So I’ll stick with a 302 for now.

Sometimes it’s fun to mix things up. It’s not as if I make any money from my blog, so I don’t mind if my search rankings drop for a while. In fact, it will be a pretty interesting experiment to see what happens with search engines and traffic.

I’m sure a bunch of stuff broke; let me know if you see anything especially horrible!

Some different browser marketshare numbers:

- Net Applications says that Chrome went from 1.15% to 1.23% in the last ~30 days.

- StatCounter says that Chrome topped 2% recently. Click through to see fewer people using Internet Explorer and more people using Firefox and Chrome over the weekends. StatCounter provides CSV export, so I made a separate chart for Chrome:

- Clicky says that in the last 60 days, Chrome has gone from 2.099% to 2.479%

Not shabby for a little over six months since Chrome was released. My favorite Chrome links recently are:

- The dev channel switcher to get the latest/greatest features in Chrome. For example, the dev channel uses the F11 key to switch to full-screen mode. You can also delete auto-form-fill suggestions by cursoring down to them and hitting the “Delete” key.

- The Chrome Experiments site demonstrates how well Chrome handles JavaScript. My favorite demos are Browser Ball, Ball Pool, Wavy Scrollbars, and the bizarrely addictive Twitch.

How do the browser stats look for your site(s)? And does anyone know of other sources for browser marketshare?

The rough summary is:
2006: 1.7M visits and 2.9M pageviews
2007: 2.3M visits and 4.8M pageviews, plus 31K RSS readers
2008: 3.4M visits and 5.7M pageviews, plus 46K RSS readers, 7986 followers on my Twitter stream, and 1607 subscribers on FriendFeed.

My most popular posts had nothing to do with search engine optimization (SEO). The top traffic-driving posts of 2008 were:
- My Gmail power tips post.
- My “Best Business Card Ever” post.
- The series of blog posts about Chrome that I did in September 2008.
- My two posts about my Halloween costume and Google’s anti-zombie robots.txt on Halloween.

In addition, my how to hack an iphone article was posted in Sept. 2007 but continued to drive especially strong traffic. If visitors were all I wanted, I’d write about nothing but the iPhone.

Almost as interesting were my traffic sources:

Google and direct visits were a large fraction of my traffic, but so were sites such as Digg, StumbleUpon, Google Image Search, Techmeme, delicious, and Twitter. It’s a good reminder that social media sites and places like image search can drive quite a bit of traffic.

All of this data is courtesy of FeedBurner and Google Analytics, which make this sort of analysis quite easy. What do your 2008 traffic stats look like?

Rough browser numbers are

OneStat says that they see 0.54% share for Google Chrome. Net Applications provides an hour-by-hour graph, which is nice, but they hardwired it to look for the string “Chrome 0.2″ when Chrome is on version 0.3 or 0.4 by now. Just eyeballing the Chrome 0.3 version stats, it looked like about 0.85% market share according to Net Applications. Hey Net Applications folks, any chance you’d be willing to roll up all the Chrome versions into your hourly report?

I hadn’t realized that Internet Explorer usage had dropped so low for my site (~26%). What does your browser marketshare stats look like for the last month or so for your site(s)?

P.S. Stephen Shankland writes about switching to Google Chrome because of the speed, while ExtremeTech also concluded that Chrome is speedy. And if you haven’t seen it, there’s a new version of Chrome (0.4.154.25) that adds a couple nice features:

Bookmark manager with import/export.
Use the ‘Customize and control Google Chrome’ (wrench) menu to open the Bookmark manager. You can search bookmarks, create folders, and drag and drop bookmarks to new locations. The Bookmark Manager’s Tools menu lets you export or import bookmarks.

Privacy section in Options.
We grouped together all of the configuration options for features that might send data to another service. Open the wrench menu, click Options, and select the Under the Hood tab.

Personally, I run the dev channel version of Chrome because I like to see what cool features are coming soon. I think the dev channel has averaged weekly updates, which is really nice because you can literally watch plug-in fixes and other improvements arrive every few days. It’s wild to see client software updated that often instead of every few months.

Update, 11/28/2008: Somehow I missed the getclicky.com browser marketshare stats from 60K+ sites. They peg Chrome at 1.55%, with a little bit of 1.6% to 1.7% in the last week or so.

“Matt, you idiot! How you could run a modern WordPress blog without caching turned on?” you may ask. The first answer is that I’m stupid and figured that I was boring enough that I wouldn’t hit digg or Slashdot for anything.

The slightly-less-stupid answer is that turning on WordPress caching interacts really badly with the FreeBSD systems that my webhost Pair runs (I’ve written about this before). That means that for me, upgrading WordPress takes hours and is a pain in the butt. The last time I upgraded my WP install, I didn’t install Supercache simply because it was such a headache to uninstall before.

I have to do some thinking about what I might change (blogging platform, webhost, etc.) in the future to make this less painful. Frankly, the idea of racing to update my software every few months (so that I don’t need to worry about getting hacked) strikes me as a little backwards; I’d prefer to host my blog with some web service so that I never have to worry about security releases or downloading/upgrading software myself. I don’t blame WordPress or Pair, but the combination doesn’t work well for me right now.

One suggestion for WordPress: integrate caching functionality more tightly so that managing caching isn’t a multi-step process and doesn’t involve juggling plug-ins. Another WordPress suggestion: lower the price of VIP hosting ($600 setup fee + $500/month is too high; if you lowered that a lot, you could probably attract a bunch of people who don’t want to fiddle with WordPress upgrades) and make it trivial to do a CNAME like blog.mattcutts.com to VIP hosting. One suggestion for WP-Cache and Supercache: make an option that clears out all cache files and deletes all directories created by the caching plugin, then atomically disables caching. Maybe that option is there and I’ve missed it somehow. One suggestion for Pair: if I am the owner of a parent directory, let me delete any file or subdirectory from that directory. Then I could delete silly cache files that are owned by “nobody.” Again, I know the fault is primarily mine for not turning on caching, even if it’s a hassle.

Since you’ve been so nice to read my self-absorbed tale of woe, here’s a picture of my cat Ozzie, helping me hack on the blog: