close dialogue

Hello smile

I'm Siobhan Curran/Kisa Naumova, and this is my weblog. I tend to write about stuff like crossdressing, Macs, code, cats, wine and Second Life, but in general it's just an ongoing conversation about all sorts of stuff. If you'd like to know a little bit more about what this all is, I recommend starting on this page which has a little bit of info on who I am, and what I'm trying to do — or you could dive into my five years worth of archives if you like.

Otherwise, feel free to close this box and explore...

Friday, 27th July, 2007

Real Estate

tagrandom mac

I love my PowerBook. He's a spritely little twelve-inch G4, rammed (pun intended) full with every configurable option when we were buying him for me off the Apple Store.

I'm not, I have to say, a great fan of Mac Book Pros. I realise that they're super speady, and a lot of people covet and worship them, but I find them unwieldy — dimension-wise anyway.

For general day-to-day text editing, email reading, web browsing, calendar stuff, and iTunes glory, my 12" PowerBook is bloody perfect. OK, so I lag like hell in SL. OK, so it takes a week to launch anything more demanding than iPhoto. OK, so the difference between encoding MPEG4s on this, rather than my G5 is measurable in hours rather than seconds...

But, I dunno. He just feels right as a machine to 'tide me over' when I'm away from my desktop.

Normally, you see, what I do is plug him into my Cinema Display when I'm at work. It's a great wee solution — akin to the PowerBook Duos of previous Apple laptop experiments¹...

(God, I'm rambling)

Basically, what I'm trying to say is that currently, all my 'big monitor working environments' are in Leeds, and I'm sat here on the sofa staring at 1024x768 pixels. Which is just too fucking teeny to get anything like what I normally do done.

¹ Remember those? Bloody marvellous they were. I still have my Duo 280c and various Docks in the loft. In fact, my first incursions into video editing were done — frame by frame — on my 280c. She was called "Baby".

Shill (LSL syntax files)

taglink bbedit lsl

Linden Scripting Language syntax colouring files for (amongst others) BBEdit

Average Colour

tagcode imagemagick awk

Just a quick one-line command to return the average colour of an image in hex, mostly for my own reference, but potentially handy for someone else. Maybe.

convert image.png -resize 1x1 txt:- | awk 'NR==2 {print $NF}'

It does, of course, rely on convert returning the colour as the last field on the second line of its output. I suppose you could use...

convert image.png  -resize 1x1 txt:- | awk '/#[A-F0-9]/ {print $NF}'

...as well.

I've got an old G3 Powerbook stuffed down the side of the bed whilst I am lazing here in bed surfing on my Dell laptop (hiss, boo, windows!)

It was strange a few years ago leaving apple behind and defecting to the 'dark side' — but I just couldn't get cheap (or the loan of!) software at the time.

I do hanker back to apple from time to time, especially recently when I was ready to update my first PC since apple — a high spec Mesh — 3 yrs old and it was groaning! :sad:

So I splashed out on a decent spec Dell with a widescreen monitor, only to discover shock horror that Vista and the graphics card wouldn't run SL. Even my crappy laptop would (albeit with major lag). So guess who's had to spend 75 quid buying a new graphics card and fitting it (and I'm not even teckie!).

Wouldn't have had that problem with a mac!

What a very sad rant!

I think I'd better go and dress as a woman to bring back some equalibrium! (no more enfemme, tgirl or giggles from me!!! — homage to yesterday!)

Have a nice day from a middle aged, grey haired, overweight bloke who has dressed as a woman since he was 6 years old.

My PB3400 sits quietly next to my really-quite-maxed-out 9500 on my desk. I still think it's one of the most ergonomic Powerbooks Apple made. Not terribly fast by current standards but it was enough to run Marathon at 25fps.

G5? iPhoto? Trying to blind me with science, dear? :smile:

Do you know what I'm still trying to get a hold of? One of these :smile:

Chatlog Prettification

tagsecondlife chat parse web

I've been struck, recently, at a couple of large(ish) discussions in Second Life, how hard it is to keep abreast of the conversation flow. The more avies in attendance, the more the topic meanders around the issues, and the more sub-threaded chats emerge.

Usually, if it's All Getting A Bit Too Much™, I shift myself into 'listening' mode, and wait until the chatlog transcript is posted at a later date.

The thing is though, it's also struck me that chatlogs themselves are incredibly hard to follow. The text is all smushed together, with very little distinction between who's saying what — and when they're saying it.

With that in mind, I've got a little project on the go that will hopefully offer an alternative 'discussion space' to the traditional 60-avies-sat-around-all-jabbering approach. It's quite a big'un, but I thought I'd share one little offshoot of it that I knocked-up (with no small amount of frustration on my afore-mentioned 12" screen) this afternoon — an Automagic Chat Log Parser And Prettificationiser...

Here's the transcript from the NMC Teachers Buzz on avatar appearance from Monday night

I'm just about to write the upload form that you use to create these — but basically you just paste or upload the chatlog, and it pulls out all the profile pictures, sorts out the formatting, makes the background the average colour of the profile pics¹ (hence the one-line command above :wink:) and — I think — makes it more of a conversation than a text-dump.

There are a few CSS tweaks to make, and I'm thinking that a choice of background colour would be good ... but I'm wondering if people think this is A Good Idea™, whether or not it would be useful, and how I might improve it.

Just one observation, regarding colours. I could have probably predicted that in general, things would tend towards 'brown' — but I'm rather chuffed with the overall colourscheme that has come out of it. One or two of my friendslist might just veer away from the beige though...

¹ Unless you're me. I hard-coded my background to be red-ish. Because I Am A Princess™

Siobhan. You can have my G3 if its any use to you. :smile:

Look on my flickr to see it

Might take you up on that...

Automagic Chat Log Parser And Prettificationiser

On the whole, I think it looks pretty good; and I'm sure it reads better than a straight log. The thing is though, it gets a bit muddy when people 'talk' across each other. Here's where I show my ignorance of these new-fangled chat thingies: Is there any was of cross referencing / linking items so that you can find answers to a specific question, without wading through the intervening chit-chat? Does the data exist in the log?

If it does, then I think it would be fairly easy to do by squirting it into a temporary database and doing a (probably backwards) pass over it to put the links in, before squirting out again all prettified.

If it doesn't, then I guess you're making the best out of what you've got. :smile:

If it doesn't

...which I'm afraid is the case :sad:

All I have is a bunch of lines of text, that are proving hard enough to regex into recongnisable patterns on their own, without trying to understand them.

The thing is though, what fascinates me is the mental processes I (or others) go through when reading it, to try and piece together the various threads of a conversation.

So, for example, say I say "I was thinking about identity and interoperable spaces" (or some such bollocks), and then a few lines later someone else says "Yes, so was I" ... what are the distinguishing things within those statements that link them together?

Is it possible that some kind of algorthymic processing can seperate out the threads in the same way that the human mind can — or is the grasp of the contexts and subtleties of meaning far to complex to reduce down to a set if if() statements?

Twitter is an interesting example of this. If I post something like "@Torley — yeah!", then Twitter recognises it as me saying something to Torley favicon, but it can't work out what I was replying to, and just links to the most recent thing that Torley has said. It makes no attempt to work out if I was replying to Torley saying "YAYZERAMA!" (which she is prone to), or "@Kisa — you like watermelons?".

I guess this is one of the things that sways me against the idea of AI — simple two-dimensional Al Gore Rythmns can sometimes seem like understanding, but they're not.

...

This is the kind of thing that keeps me awake at night. I hope you can understand why I drink so much in compensation now :unsure:

...

Just one thing that's bugging me about this though. SL profile picture tend to be of an avie in a particular place (unless you're me, and have tried to be artsy). So the pictures that accompany the chat place the 'speaker' outside of the place where they were talking when the discussion was going on.

I'm not overly fussed about that — I mean, I'm curious about the difference between a profile picture and an avatar/buddyicon, obviously — what's really bugging me is that it's reminding me somewhat of some part of a children's TV programme. Something where people would talk in a conversation, but the photo accompanying what they were saying where from completely different spaces.

I'm guessing that no-one has any idea what I'm talking about :unsure:

...far to complex to reduce down to a set if [sic] if() statements?

Yes, it is too complex for software. Wetware can do it because it has massive, parallel, resources that have evolved over millennia.

I hope you can understand why I drink so much in compensation now

No. If you had any common sense, you'd invoke the "fuck it, it's too hard" rule long before becoming so obsessed you need to anaesthetise yourself in order to sleep. :tongue:

(That last bit comes across way harsher than I really intend — the tongue is more in the cheek than sticking out :biggrin: )

If you had any common sense

Oh come on — this is me we're talking about :wink:

Hi Graham, just to let you know Beckka and Podge are up this weekend, might do something later if you are around give us a call.

gravatar

Jon

Sorry Jon, I was lying on my back in bed when you rang earlier, trying to compose a post all about how I've done my back in somehow. Call you in a bit.

Turing Machine. With a difference: you want it to interpret as a human, instead of providing responses that might be indistinguishable from a machine.

You might be waiting a long time. Not even the best Neural Networks are up to that sort of pattern matching — yet. (NN research seems to have stagnated over the last few years. There was a mighty push about 8 years ago, and then... Not a heck of a lot of progress.

C'est la vie

Carolyn Ann

er "indistinguishable from a human.

Sorry.

Carolyn Ann