Log in

No account? Create an account

Being a Twit

Playing with twitter, but half my entries are "updating my twitter feed".

Right now it amuses me to think that my sense of humour is calibrated so as to make it impossible for people to tell when I begin to go senile (on the off chance I live that long). In continuance of this, I offer the following rationale:

  • Go to twitter site
  • Spend 3 minutes clicking on my own twitter profile before I again figure out that I need to go to another page to actually make a new entry
  • Get confused/distracted by something else/other idea pops in to head and needs handling
  • Spawn brain-thread saying "what am I doing?"
  • Page loads with a nice big text box
  • (background thread places in cache "You are updating your twitter feed. Twitter is a site for you to log what you're doing")
  • New directive accepted, prepare to log what you're doing
  • Spawn brain-thread saying "what am I doing?"
  • Get cached result "You are updating your twitter feed"
  • You-are-being-ridiculous LED lights up, but there has been tape over that for many years
  • Type "Updating my twitter feed" into text box, hit update
To what extent this is senility is left to the reader.

Mental state has been bleak this weekend, spent most of saturday asleep dreaming about skiing on utterly, creepily empty hills and empty ski lodges, most of today thinking about things for work and delivering some food to someone who wasn't feeling well. Cats are extra snuggly as the cold and darkness push inwards on apartment windows.

While I've finally found the magic settings to make the "ink" tool look awesome with my writing (for webcomic), I realise now that the versions I actually put up are too small for most of it to be seen. This suggests metaphors - what proportion of what's actually going on in the authors/philosophers/poets/musicians/etc that we like actually make it to us? Do we get the best bits? If someone put a lot of thought into the fine details of something but didn't feel like explaining the basics, we might see nothing of their labour - could we ask a musician to justify the use of some musicial component in a composition rather than another? Maybe sometimes the answer would be "because it felt right", but perhaps sometimes it could be a pleasant day's conversation, or perhaps they actually will have directed/composed alternate forms of the song with that component switched. Meaning is like that..

Random additional thought about the Socialist Equality Party - I think I would rather have a Democrat in power than the candidate (or many of the people in the room) I met there. It is great that he actually could intelligently discuss the questions I brought up about the British Labour party, the CCP, and entryism, and that he appeared to have a reasonably good knowledge of history, but his analyses were so strongly dogmatic and orthodox as to make him effectively stupid in handling the challenges of running a nation (or reforming one into the stages past capitalism). I doubt a state that had such a party in power could accomplish much positive in the long run - perhaps people with too dogmatic a view make necessary the doublethink that characterised and stifled Soviet (and to a lesser extent East German) life, as "a priori" dogma refuses to be made practical but practical things must be done anyhow to keep the state going. I suspect this is the case wherever sufficiently strong/inflexible/broad dogma (rather than strong values with some flexibility in dogma) guides a state.

FreeBSD still makes me want to scream in how staggeringly backwards it is in a number of important bits (some basic technology, some user interface stuff). Linuux has never been (and is not) a particularly good Unix, but...

All this has me thinking about the device naming conventions - I can't decide if I like devices getting names from the driver they're handled by or by the device type they are. It is kind of nice in some ways that for the project I'm on at work, the wired ethernet devices have predictable names under FreeBSD thanks to them having driver-names, but at the same time it really sucks that system management stuff needs to change for any node where we swap out one piece of hardware for another - it might be nice to have eth1 stay eth1 rather than msk0 become fnord0. Regarding slices, partitions, and all that other fun stuff, I really don't know what's better, but it's a pain in the butt that for an image of a raw disk, Linux can't (despite futzing with virtual-device interfaces, fdisk, and offsets to "mount") seem to manage nicely handling BSD partitions that live inside a BSD container on there.

I suppose offering multiple views of devices simultaneously might be nice - it would be pretty nice to be able to have your FnordComIII ethernet adapter available as both /dev/lan/fnord0 and /dev/lan/eth2

I've also come around to thinking that devfs is in fact a good idea, having worked with images and virtualising devices and the like enough (and maybe having thought about the cool things Plan9 and GRUB does). It'd be great to have utilities that would let one manage layering of the network stack on a raw device (optionally letting people also associating the lower-level devices for when violating abstraction leads to big performance wins), particularly for layer2 and layer3 tunneling. I'd love to see this applied to filesystems as well, with whatever OS is being used handling, with no awkwardness or nonstandard tools, Linux, BSD, or Mac-style partition tables.

If only Plan9 had had less brain-damage..


I would expect Linux to be able to handle your desired mount-scenario. When you say "offsets to mount", you mean you manually forced mount to see the start of the slice within the image, and not the start of the image? Do slices work like partitions once you get inside them -- just an image of a filesystem? Presuming that's the case, it should _really_ be possible just to index into the image and mount from there.
In theory, slices do in fact act like partitions that way - I was a bit surprised that it didn't work to give an offset (although perhaps mount_ufs is a bit broken or fragile compared to the other mount backends...). Of course, in an ideal world I wouldn't have to do such a hack to find the partition (and it is at least possible that the multiple layers of indirection needed to dig through the image-layer, and the BSD partition layer to reach the specific slice I was trying to mount caused me to get the math wrong, or alternatively that the ufs driver can't handle FreeBSD's ufs at the moment - the error messages were not at all helpful).

All this stuff is far more difficult than it should be - the userland and kernel stuff for a Linux distro for a different architecture (e.g. MacPPC) might be more intelligent about it. It's a pity I don't still have my old iBook.
When I used to work with OpenBSD, I don't think I had a problem mounting BSD slices from Linux, as long as I had the Linux kernel's BSD partitioning option turned on.

devfs is dead since it didn't really work with SMP systems; udev is the equivalent now. And I love that udev allows me to get to a disk partition with either /dev/sda1 or /dev/disk/by-id/usb-WDC_WD25_00BB-00GUC0_1144-0:0-part1. Though occasionally that 0:0 changes to something else, or disappears entirely, which is annoying.

I like your comments about flexibility vs dogma.

I thought it was easy just to click on "home" at the top of the twitter site....
By devfs, I mean the generic idea, not the specific implementation in Linux.

I think for bare device names, I really like the Solaris naming conventions..

I just need to remember that I *do* need to click "home" - if I start to use twitter more frequently and don't just update my blog/wiki software to post for me, it's going to continually confuse me, I think :)
I agree about Solaris names, for fixed interfaces. For dynamic interfaces like USB it'd be more of a pain though.

If you start to use twitter more frequently, just bookmark http://twitter.com/home and you're set, or just leave that page up in a tab all the time.