How to convert a FAT32 disk to NTFS

I had a near-Windows experience today.

I have had an old 500GB USB disk sitting around, that I used for manual backups in pre-Time Machine times, back when I deemed FAT32 to be the comfy file system choice for hooking up a disk with any computer I might want to reach it from. Today, I think that sweet spot is NTFS.

Especially as Time Machine doesn't want anything to do with FAT32 devices as backup targets.

I figured the easiest way to convert the filesystem in place (as I didn't have any other disk around) was to boot up a Windows XP virtual machine in Parallels and let Windows do it.

That procedure was perilous. If you ever embark on doing it yourself, here is how you probably want to do it:

> net stop wuauserv
The Automatic Updates service is stopping....
The Automatic Updates service was stopped successfully.

> convert E: /fs:ntfs
The type of the file system is FAT32.
Enter current volume label for drive E: [...]

> net start wuauserv
The Automatic Updates service is starting.
The Automatic Updates service was started successfully.

If you skip that first step, you are wise to remember that a Windows machine needs baby sitting not to do bad things to itself. Like deciding it has found and applied security updates that it really wants to restart to complete in ten, nine, eight, ...all the while that long file system conversion pass is humming about in the background.

Fixing $(document).ready() to work for any document

The first time I saw jQuery's $(document).ready(callback) idiom, I thought it was a great idea: not only is it readable as plain English, grep(1)able (it is available as $(callback) too, but I advise against using that in larger code bases, where you might need to dig up all references to it, at some time) and intuitive, it also lets you provide a different document object you are waiting for.

The first time I used it in the latter capacity I was a bit saddened to find that it was not quite that good yet; it always tests the same document that jQuery itself was loaded into. (At present time, meaning jQuery 1.3.2 and earlier.) The document object you pass isn't actually even fed to all the code that together implements the ready function.

As we needed it to do that at MashLogic though, my colleague Dave took to fixing it so it does indeed work that way, as per the jQuery ticket that has been sitting around for two years (from someone else that apparently wanted and expected the same thing), and I submitted a patch upstream. I'm hopeful it gets accepted for next jQuery release.

We even prepared a little test case that loads jQuery in an iframe and puts document ready callbacks on both itself and its parent window, which seems to work fine in all browsers and OSes we have tested it in so far.

I think the original ticket may have gotten closed over it often being a bit tricky to get a reference to the document object of an arbitrary frame or iframe for some frame you want to track -- the javascript security model forbidding DOM access to content in documents loaded from a different domain, for instance, and timing issues with grabbing a reference to the document object of a frame that is inlined in a document's source -- before we can get a handle to the document, it is a little difficult to have that parameter to pass to jQuery for listening in the first place, after all. But I hope such issues don't prevent the fix from landing, for the cases when we do have one.

Tracking multiple svn branches in git

I've been using git-svn for a while, to track projects still trapped in the subversion age, with git. It's not too messy to set up, though in order to get nice names addresses you need to map all the subversion identities to their corresponding name and email address. I recommend taking the time to suck out the entire repository history -- it's a one-time operation, and you get free offline history browsing and inspection.

If you have the names and email addresses of all svn committers, list them in a newline-separated file, each line reading svn-id = Person's Name <email@address> -- I kind of started tracking a list of jQuery committers, but got bored and/or side-tracked before finding the last ones. Anyway, if you go through the trouble, pass along this file on the command line (git svn clone -A committers.txt ...) when you create your local copy.

The docs for git-svn are not the best, and didn't mention how to import news into your local git repository for more than the branch you have checked out at the moment. The trick, which might be obvious for some but was not for me (thanks, singpolyma!), is to run git svn fetch --all to update your local repository with data for all branches, instead of (well, or besides) git svn rebase that just updates your currently checked-out branch.

And Bob's your uncle.