Raw

The days are getting longer, sap's rising and I'm in a positive frame of mind so seems a good time to jot down what I'm up to.

Personal Cruft

A few months ago I went on another bit of a down, and predictably fell back on my usual (lousy) coping strategy, drinking loads of wine. I wasn't working, so have drained all available money. But I've pretty much pulled out of that patch now, regained a bit of optimism and am currently on the wagon. Still living in rural Italy on my own aside from 2 dogs, 2 cats, though have got romantically involved with someone, albeit (very) remotely so far.

Work-Work

The past couple of months I've been coding on Apache Stanbol, getting paid indirectly through the Fusepool project. My role there is near-enough in line with my long-term plan of creating a loose consultancy. I'm only now getting anything like up to speed on Stanbol, but it's interesting material and when I actually get stuff done, pays ok. It'll still be a while before I can relax, credit card is still maxed out etc. but at least things seem to be going in the right direction.

Coding

I've neglected my own code for a long while, first just not being in the mood and then having to concentrate on work-work. A big problem I have frequently had is having too many balls in the air meaning nothing ever gets finished. I've been using Workflowy for (hierarchical) todo lists for a while, but the number of items on the lists tends to increase much faster than I get things done. So I'm going to limit myself to about 4.5 projects, and cycle through them, just making an approx. one week to do list for each on each cycle. The projects are:

• Seki (server-side node.js) - a bit of linked data middleware
• Scute (desktop Java) - a Semantic Web hacking utility
• JEdwards (desktop Java) - a custom text editor optimised for working on Javascript/node.js, is a sub-project of Scute
• Web Beep (server-side Java) - text to tones and back again
• Site - anything to do with my online presence (admin, blog, docs, services, vocabs etc.)

Writing

No professional writing in the pipeline, I was finding book writing a bit too stressful for the benefit it provided (last thing was working on a revision of Beginning XML). Got a few ideas for online tech pieces, longer than typical blog posts, that I'd like to assemble as a kind-of book. But there are no deadlines (or payment) associated with them. Still haven't started the novel...

New Place

Although it was on hold over winter, I'm still in the process of renovating the house opposite. The other day I took a load of photos that I put into a presentation format as a visual todo list (the text items the pasted into Workflowy - data reuse!). I've said several times that I plan to be in there by xmas, but now it actually seems within reach. The biggest job, which had become a bit of an albatross, making the windows, is getting close to completion. None of the other things really prevent me from moving in, so now I can realistically say "in by summer".

Other Stuff

Way too much time on Facebook, at the cost of blogging, Twitter and G+. I've continued to read a lot of tech stuff but haven't been interacting on any mailing lists or working groups (in fact I got bumped from the HTML5 group because the renew form didn't work for me - I told the given contact but was ignored, well bugger 'em). That's liberated a lot of time :)

I have been doing a little bit of woodcarving, music and electronics. No travel plans, but all being well a certain someone will be coming over here from Australia very soon.

A few weeks ago I read an article about making useful audio level measurements. It occurred to me at the time that it'd be pretty straightforward to set up the recommended approaches in software (as a VST), but would also be nice to have a little dedicated unit in the studio. Easiest way would be to cobble together the appropriate analog circuits, though finding a decent true RMS converter might take a while. Then yesterday I was going through some boxes on my mother's landing and came across a magazine article I'd written in 1994 on, guess what, RMS converters. So I've scanned it and put it online : Reading RMS (pdf of scanned images, only 7 pages but 25MB!). Some other bits I did for WW are linked from Chaotic Electronis and Bats.

Last night I posted an email to list objecting strongly to a post by Hixie. I broke protocol. I ranted, took it personal, I basically called him a fascist, no doubt swore a bit too. But unlike many previous occasions when I've got on my high horse about stuff after a few glasses of wine, no shame this morning, I'm quite pleased I went that far, it really has got beyond pleasantries. Trying on a one-man coup as acceptable behaviour makes my sweariness pale into insignificance.

I immediately got kick-back from danbri, Bijan and the other (list gov) Ian. Danbri made the point that although Hixie might have been into eugenics in the past, he should be given the space to change. Fine - but I see no counter-arguments from those beliefs in recent times. Couple those beliefs with the evidence of a wish to take sole control... danbri quoted Godwin's law at me. In response, I have to say: if it quacks like a duck and, er, steps like a goose.

So onto the meat:

In the first paragraph, Ian associates consensus with the perjorative term "design by committee". That's just disingenous. For standards to even remotely work for all interested parties, those parties need to have some power over decisions. It will be always be a suboptimal compromise, but...well wait, he has an alternative.

One person. I wonder who he might have in mind.

He makes a fair point with "consensus amongst those who bothered to show up", but personally I believe that's a positive, the people that turn up have some kind of commitment to the matter at hand. In fact, here is he not saying "I'm prepared to show up enough that I can take this all off your hands".

Personally I also agree that "[specs] should be designed so that they end up solving real problems" - but isn't that a redundant statement? Why bother if they didn't?

I can see where he's coming from with: "I need to make sure that whatever I spec is something that the bulk of implementors want to implement, otherwise it goes nowhere." The only problem I see with that is, well the "I".

Please don't get me wrong, I do have huge respect for Hixie's brain, for technical stuff I feel C-list in comparison. But when it comes to big picture stuff, I probably have a slight advantage over many folks that spend their days with noses in code. I've seen political machinations in many other contexts, I know an attempted land grab when I see it.

But ok, assuming Hixies motivation is noble (is clearly committed, but I'm suggesting misguided) - there's an easy test. Would he be prepared to stand down and allow someone else to take that place? (noting there are plenty of capable candidates in the WHATWG)

(Comments to G+ please)

Around 1992 I'd been reading about Chaos theory and started playing around with some analog electronic circuits. I was amazed how easy it was to build a chaotic circuit, and on impulse I sent a couple of my designs off to Wireless World (the magazine best known for publishing Arthur C. Clarke's idea of geostationary communication satellites). More amazement, they published them in their Circuit Ideas section of June of that year. My first published work. The editor asked me if I had any more similar material. Well it took me a while, but I did a full article - The Twisted World of Non-Linear Electronics (PDF of images scans from mag, 9 pages - includes the chaotic stuff, which was republished in a little Newnes book). To my delight, it was the cover article in the February 1993 issue, and for the cover they did a brilliant modified version of Dali's "Persistence of Vision", with melting circuit boards instead of clocks. I think the only other thing I did for them was a piece on Operational Transconductance Amplifiers, which sound pretty obscure but are actually really nifty. The most notable circuit in that piece, The Versatile World of OTAs (March 1994 issue, PDF, 5 pages), was a heterodyne ultrasonic receiver - in other words, a bat detector :)

I'm pretty sure the material's all still valid (even though analog). A while back I hunted for old WW stuff on the WWW, alas nowhere to be found. Fortunately I'd "filed" my only copies of the magazines at my mother's, remarkably she was able to find them, picked them up last visit. Only just got around to scanning them.

We now know how to do this computer stuff. Networks even. The Web of Data is now mainstream, yesterday's news. But we're still at the stage like Luther pinning his notice to the church door, can't find a stapler. Out of my window I see a little electricity pylon, that will be on a DB somewhere. I also can probably see 50 trees. I should be able to describe them individually, that's actually a plum tree that I cropped last year. I should be able to identify the caterpillar crawling up the twig, on that branch, of that tree. It's totally Engelbart, augment the person. But I don't think anyone (except perhaps for Phillip K. Dick) envisaged the possibility that this could be networked. The separation between human individuals can be huge, but there is a way of having a shared intellectual overlay of the whole of reality. We have all the bits of technology, easy, already. To pick a cliche, imagine there's a starving kid in Africa. We can know about that almost immediately, divert appropriate resources. Ok, just some guy with a mobile, taking a piece of bread. As a species, we have the potential to be so beautiful. First time on this this planet, let's make the best of it.

There is now a critical mass of people that know about Web data. Call it semweb, whatever. You can see some frustration bleed out, language in Facebook post (alright, I still just want a taxonomy for prostitutes). The ideas of 10 years ago have been fulfilled. We have the Web of Data, in a year or two, the momentum is rolling, different than imagined, but it is here. Now we have to look at other interesting things, that might be useful for humanity. I'm not sure, but anonymous access to the Web seems a good idea. Protocol work. Next.

Main thing, I'm now available for work. Ideally coding/related or writing, part-time, but anything considered. It has to be remote working, I've too many commitments here for anything other than occasional travel.

So short term I'm just up for anything that'll help pay the bills. Longer term, my plan is to operate under the umbrella of hyperdata.org. I'm labeling it a consultancy, with primary focus around Semantic Web technologies, but that's only a first approximation of what I've got in mind. Yes, I'll be available as a consultant (and for hands-on coding etc), but there will be two key differences from most traditional consultancies.

The first is that if I'm not convinced I personally can handle the job in question, I know a lot of people working in the related fields. I'm confident I can find appropriate expertise. So in that sense, hyperdata.org will operate as a kind of agency (initially I intend to organise this very loosely, informally, but if it seems appropriate a more formal setup is always an option). [If you've got free cycles let me know.]

The other novel aspect will be the approach to research & development. Many tech consultancies have ongoing dev projects, but they tend to be very focused around the core of their consultancy work. In the extreme (but common) case is that a particular piece or set of software is at the heart of the business, the consultancy in effect being an extended kind of support for that software.

I want to go in a different direction, actively trying to avoid any predetermined path, rather taking a more exploratory approach. The idea is to do a lot of "spike"-style development - comparatively agile and rapid, in general only taking things to the proof-of-concept stage. My little seed (and feasibility check) for this was Web Beeps. A small but fairly novel idea, which I developed just enough to get a live service running, and which in turn has spawned a load more ideas for experimentation. Web Beeps took me about a month part-time to take from notion to service, and that's the kind of ballpark timescale I've got in mind for discrete research projects.

The primary intention behind this R&D is to inform other activities, to discover effective techniques and processes. Think hobby coding or Googlesque 20% time, but on the one hand with a slightly more formal approach (a clear, typically finite life cycle for projects), on the other hand with no fixed time/resources allocation - depending on other work, it can vary between 0 and 100% of work time. If there's a contract to fulfil, that takes total precedence.

A side effect may be that certain spikes suggest themselves as suitable for further development towards some kind of commercial product. In such a case, my current feeling is that it will probably be preferable to spin them off to third parties at the first opportunity. Time will tell how that pans out. As I'll mostly be working on my own, I do have the huge advantage of being able to be flexible about business model, so if any avenues do look particularly promising I can change practice to suit.

On a personal note, experience suggests I'm most productive at computer-oriented work when I'm putting in about half the hours of a typical full-time job. Given my current circumstances, that's actually a positive. I've no great ambition to get rich quick, the motivation is just to pay the bills doing something interesting. My cost of living is relatively low, which helps. Also I'm still renovating a house, and have got loads of other hobbies, so restricted hours is a good fit.

If you think I may be of use to you, let me know.

Main thing, I'm now available for work. Ideally coding/related or writing, part-time, but anything considered. It has to be remote working, I've too many commitments here for anything other than occasional travel.

So short term I'm just up for anything that'll help pay the bills. Longer term, my plan is to operate under the umbrella of hyperdata.org. I'm labeling it a consultancy, with primary focus around Semantic Web technologies, but that's only a first approximation of what I've got in mind. Yes, I'll be available as a consultant (and for hands-on coding etc), but there will be two key differences from most traditional consultancies.

The first is that if I'm not convinced I personally can handle the job in question, I know a lot of people working in the related fields. I'm confident I can find appropriate expertise. So in that sense, hyperdata.org will operate as a kind of agency (initially I intend to organise this very loosely, informally, but if it seems appropriate a more formal setup is always an option). [If you've got free cycles let me know.]

The other novel aspect will be the approach to research & development. Many tech consultancies have ongoing dev projects, but they tend to be very focused around the core of their consultancy work. In the extreme (but common) case is that a particular piece or set of software is at the heart of the business, the consultancy in effect being an extended kind of support for that software.

I want to go in a different direction, actively trying to avoid any predetermined path, rather taking a more exploratory approach. The idea is to do a lot of "spike"-style development - comparatively agile and rapid, in general only taking things to the proof-of-concept stage. My little seed (and feasibility check) for this was Web Beeps. A small but fairly novel idea, which I developed just enough to get a live service running, and which in turn has spawned a load more ideas for experimentation. Web Beeps took me about a month part-time to take from notion to service, and that's the kind of ballpark timescale I've got in mind for discrete research projects.

The primary intention behind this R&D is to inform other activities, to discover effective techniques and processes. Think hobby coding or Googlesque 20% time, but on the one hand with a slightly more formal approach (a clear, typically finite life cycle for projects), on the other hand with no fixed time/resources allocation - depending on other work, it can vary between 0 and 100% of work time. If there's a contract to fulfil, that takes total precedence.

A side effect may be that certain spikes suggest themselves as suitable for further development towards some kind of commercial product. In such a case, my current feeling is that it will probably be preferable to spin them off to third parties at the first opportunity. Time will tell how that pans out. As I'll mostly be working on my own, I do have the huge advantage of being able to be flexible about business model, so if any avenues do look particularly promising I can change practice to suit.

On a personal note, experience suggests I'm most productive at computer-oriented work when I'm putting in about half the hours of a typical full-time job. Given my current circumstances, that's actually a positive. I've no great ambition to get rich quick, the motivation is just to pay the bills doing something interesting. My cost of living is relatively low, which helps. Also I'm still renovating a house, and have got loads of other hobbies, so restricted hours is a good fit.

If you think I may be of use to you, let me know.

WebGL appears to be disabled by default, but for me at least it was possible to get it going by adding the following arguments to the launcher in the apps menu -

--enable-webgl --allow-file-access-from-files --ignore-gpu-blacklist

My Intel/Mesa card appears to be blacklisted, open issue, but so far it seems to work ok.

Looking at chrome://gpu in Chrome I now see:

Graphics Feature Status
Canvas: Hardware accelerated
Compositing: Hardware accelerated
3D CSS: Hardware accelerated
CSS Animation: Software animated.
WebGL: Hardware accelerated
WebGL multisampling: Hardware accelerated

Problems Detected
Accelerated CSS animation has been disabled at the command line.

No idea why the CSS acceleration is disabled.

There are also a load of things to tweak under chrome://flags. (There are a huge number of other internal URIs, see chrome://about).

Some relevant refs. I came across: WebGL in Chrome 10 on Ubuntu 10.10 (current is more like Chrome 20, Ubuntu 12.04, but this all seems to still be valid), List of Chromium Command Line Switches, Chrome Experiments.

I'm still a bit mystified by this next thing. After upgrading Ubuntu, I noticed in Ubuntu, System Settings -> Graphics I have:

Driver : Mobile Intel® GM45 Express Chipset x86/MMX/SSE2

Experience : Fallback

Which suggests a fallback driver is in use, yet:

~$ glxinfo | grep render

direct rendering: Yes

OpenGL renderer string: Mesa DRI Mobile Intel® GM45 Express Chipset x86/MMX/SSE2

GL_EXT_vertex_array_bgra, GL_NV_conditional_render,

suggests it's using the correct driver...

Comments to G+ please.

I just ran the upgrade from Oneiric Ocelot 11.10 to Precise Pangolin 12.04 using the (GUI) upgrade tool. On reboot I got a garbled screen. There could be many causes for this, but in my case it was because the installation failed silently. Luckily the first thing I tried led me to the solution:

Opened a shell window -

Ctrl+Alt+F1

ran -

apt-get update

tried -

apt-get upgrade

it gave an error, so ran -

apt-get -f install

another error, referring to a blank line in /var/lib/dpkg/status, in the Description field for xmind. I removed the blank line here, then afterwards in /var/lib/dpkg/available, then again ran -

apt-get -f install

this time it ran to completion. On reboot everything appears to be ok, and it's even kept my desktop settings (Gnome Classic, I didn't get on with Unity).

My guess is the blank line in the package description of xmind led it to fail in the first place. I'll ping them.