It’s in some degree a little tragic that holidays give me a chance to REALLY GEEK OUT – as though I don’t have endless opportunities at other times. But, honestly, and perhaps equally as tragically, geeking out is what I do when I need to become philosophical and retrospective and moody and irritated at everyone. It also encourages me to drink heavily, all of which I think are the true meanings of festivals in the human condition anyway.
My geeking out this Thanksgiving (apart from a hard-to-fight urge to buy consumer electronics, which I swear to God must be down to morphic resonance, because I’m really not that interested in retail usually, have successfully ad-blocked most of my life, and don’t have any money right now anyway), was a day zero purge of my servers and laptop home directories.
Usually this kind of thing ends up in complete disaster, like deciding one day to take your car apart and put it back together again.
This time it’s been working out rather well. Instead of naked and crying on the floor, I’m clear-headed though a bit chilly.
There wasn’t really any reason for it, except that recently I’ve been very forward-looking: as though I’ve been trembling on the edge of a precipice in a wing-suit. I’ve not been looking down, and I’ve not been jumping, but I’ve certainly not been looking behind me either. So it was about time to push myself over.
rm -rf! goes the urge, and I gave into it. I’m slowly re-introducing (or rewriting) all the scripts that I use, and taking into account the lessons I’ve learnt in the last few years.
It’s really easy to be frozen in the headlights when you delete your existing directory structure and start again, because you end up thinking so much about the future. But you can’t tell anything about the future, so there’s no point to being frozen. You just have to express confidence that it’s not going to be awful, and jump.
I have a fairly concrete aim, which is to see how close I can get to having a setup that is a) replicated everywhere, and where I can b) fall-back to different machines if one breaks, and c) I can throw at an EC2 instance as easily as I can throw it at a Nokia N810 or my Mac laptop, and d) shares as much as I can to the rest of the world.
God knows, I won’t get there, but as this has been my aim for some years, I have some lessons learnt, and some new technology has been rolling along for it. For instance, for years I’ve been following Joey Hess’s Living Life In Subversion credo: version-controlling my home directory, and trying to keep as much of it as public as I could.
It’s a great way to think of your digital life, not because of the fact that it keeps all your documents as revertable backups (like MacOS’s Time Machine) or allows you to sync your home directory across many systems so much, as the discipline of thinking “how much on my computer should be private, and how much should be public?”. Joey keeps a huge chunk of his home directory in a public repository, and it’s incredibly educational – both for readers and for him, I suspect.
Being more public is terrifying, and yet freeing at the same time: apart from anything else, you quickly learn to discriminate between public-because-i-created-it-and-want-to-share and public-because-it’s-not-actually-mine. That’s to say, I have a bunch of free ebooks, say, that I can make public because they’re public documents. But at the same time, I don’t worry too much about backing them up, because I have a world of backups out there already. I’m sure those of you who torrent films or used to file-share music will recognise that feeling. Why should I keep this, when so many others have a copy I could obtain easily enough? My copy of Brian Eno’s albums takes up a few megabytes on my hard drive. Should I back it up? Or should I just keep the receipt, knowing I could get a new copy so quickly if it was lost locally, here?
Thinking about what’s really private is also very clarifying. Passwords are private. Are bookmarks? Which bookmarks? What’s the minimum set of bookmarks I can make private? More people are asking this after their heavy del.icio.us use, I bet.
Here’s my current new directory structure. Like the old one, the public bits will appear sooner or later as a browseable repository on this machine. The private stuff is ghettoised into a single folder – directories that hold things like ssh settings are symlinked into it.
|
1 2 3 4 5 6 7 8 9 |
meta # scripts and Changelogs that I wrote to track all of this private # projects with data I don't want the world to know public # everything else, listed by project directories mail # Archives in mailbox, current in Maildir src # symlinks to any folder called "src" below private, meta or public bin # a nest of symlinks to files in folders claled "bin" in private, meta or public web # websites, including this one tmp # anything older than ten days in this is deleted. It's where I save files. logs # growing files (not many: I'm not going to default to logging this time) |
Most of this is (now) kept under Mercurial, a distributed version control system. The scripts are in Python, where I can help it. The structure is replicated across all my machines, with the same contents. Not everything is amenable to version control, but I have some ideas about how to keep the other stuff mirrored across all my machines too.
I’ve also got all my machines talking to each over IPV6. I want them to become more chatty, and less like they’re hiding on the edge. And I’m also pushing the edges of designing this system so I can share it with my closest friends: have it flexible enough to know when it is being used by someone else.
I realise some of this must not make much sense, but hopefully as I explore more, what I’m trying to do will become clearer. I’ll get all seasonal and aphoristic until I get there: Doing this kind of purge isn’t my way of apologising for the past, but thanking the future in advance.