18 August 2019
A quiet week spent exploring with my hands.
I continued my explorations of Dokku as a hosting option, spending some time looking at best-practice for deploying Wordpress to it - and then working out how that lined up with my own preferences around deployment (notably, specifying Wordpress as a dependency in composer).
That was a useful exploration - getting hand dirty with some servers, turning something I’d often end up doing by hand into something more automated - and I at least got to something sensible and working by the end of the week.
I spent a day documenting what I’d done. Not ‘documentation’ that resides in the text files I use as a digital notebook; ‘documentation’ other people would find useful. That’s a challenging thing to write clearly and unambiguously. The end result was perhaps too long, but I think it explains itself clearly enough, and includes appropriate code snippets and samples.
I hope I’ll publish that somewhere next week. If nothing else, it was good practice to write some real documentation for something technical, multi-stage, and complex. I’ve got more writing coming up this month, and it was good to get a handle on my pace of writing (and, for technical work, getting a feel for how often I need to bounce between edits and writing). And, who knows, it may still be useful for someone.
On Friday, my circuit boards for Dent arrived from China and I set about assembly. These are the smallest component sets I’ve ever had to assemble - the 0603 is all fine, but the QFN microcontroller has been a pain. (QFN is probably the first of the surface-mount IC packages that is ultra-hard to do by hand, as it has no exposed pins. It’s really designed for a robot and oven to do).
I’ve been practicing with solder paste, stencils, and hot air, and whilst the first stage of my assembly seemed to go OK, the usb socket on the board has possibly not gone in correctly, leading to some powering issues, a very hot chip, and a frustrated me to end the week with.
Still, I have an idea where to take it next week, and I solved a few problems at the firmware/bootloader end for once I’ve got the chip up and running.
Quiet, but busy, and some good studio conversations with my colleagues throughout.
11 August 2019
After an intense week 343, 344 was much quieter. I got some feedback on the work so far on Longridge, so spent another afternoon or so wrapping up my work there, going over that feedback, adding some new things, and sending it all off for approval.
A box of parts arrived for Dent. Not much happening there til the prototype boards arrive, however, so that’s sat on my desk.
In “learning things” time, I spent a little while exploring new deployment options for servers. In particular, I spent some time looking at Dokku, a containerised deployment platform that installs to a greenfield server environment and then allows deploys via git pushes. It uses Heroku-style ‘buildpacks’ to provision and configure server infrastructure. I’m not sure I’ll move to it just yet, but a small amount of work got me a setup that will compile Hugo sites upon deployment and then point nginx at the deployed folder. Useful to consider as a way of gardening future servers, and also of standardising deployments and tools for myself.
And at the weekend, I saw in this picture of Richie Hawtin playing live… that he’s got a 16n as part of his rig (spot the faders between the tangle of cables that is his modular synthesizer, and the little 303-clones by his waist):
I was pleased and proud of that.
4 August 2019
A busy week: lots of work, not a huge amount to say, as a result.
I spent four days at After the flood on High Vinnalls. This was a product development and data exploration exercise. I worked on exploring some datasets for an ongoing client project over theirs, building tools to quickly spelunk around in the data and establish signal, noise, and see what other things would reveal themselves. Not in order to visualise them, necessarily, nor to make a data-exploration product. Rather, the exploration was to support product development and invention. What products might be possible? Does the data support various ambitions?
ATF presented to the client at the end of the week, so it was an intense few days of thinking, sketching, and coding. Really good to be in the room working closely with designers as we tried to understand the shape of what we’d been given to work with.
In amongst that, I went up to Leeds on Wednesday night to spend all day THursday on a workshop at the University for Longridge. The goal of the workshop was to devise the structure and outline of three short courses I’d been writing. This too was very, very intense; I think we made good progress, but I’ll need to return to what we did in week 344 to finish up a few last things - and to be able to look back on it with the benefits of some perspective!
I also got a goahead at the weekend for an initial exploration into Dent, which may, or may not, turn into a project in due course. For now, it’s a quick pass to just see if anything is possible. This is a small hardware project - very much something on the side, but a nice backburner project if it comes off.
Phew. Busy, busy week.
28 July 2019
Longridge is really beginning to motor. I finished writing a the initial preparatory work for a workshop in Week 343. That was harder than planned. Lots of variables are still up in the air, and the later stages of the work are highly dependent on these ones; as a result, I sometimes get tangled trying to think about all the possible outcomes. But: I got enough done to generate the raw materials I was required to for the workshop, which was most important.
Also, the topic area is now well established around in my head, which means I have lots more thoughts still unwritten - some aren’t quite ready for that, others might emerge in the right context. I think that’s all fine: it’s the right place to be in at this point, and will be more things to feed in to the workshop next week.
Towards the end of the week, a short piece of work for After The Flood came in. I’m going to be spending the rest of Week 343 with them, working on some product exploration around data for them - coding, talking, thinking work. Looking forward to working with that team again.
I also have a small hardware project that has had some external encouragement to suggest it’s worth getting to the end of phase zero on it. Phase zero is, I suppose, the point where something reveals if it’s going to be A Thing or not. Sometimes, you do the work, and there’s just not a product there (or not one worth building); better to find that out with as little effort as possible. I’m going to finish up this exploration phase and see where it lands. For now, that means getting some PCBs spun and assembling them when they arrive.
My little React prototype is in good shape. Something clicked and I tore out lots of component-spaghetti. Now there are just enough components. All of a sudden, passing state down looks tidy, rather than baroque. I spent some time adding a few little features, and continuing to refine my knowledge of new browser APIs. It still feels like a delight to find so many browser APIs being generally supported - I come from an era where most of the Good Stuff just wans’t standard enough. And, on top of it: the little tool is beginning to feel usable.
And that’s about it. Week 343 is a busy one: in Leeds on Thursday for Longridge, and at ATF for the rest of the week.
20 July 2019
I kicked off work on Longridge this week, with a remote meeting to go over the shape of the initial work, and some of the approach. That’s given me some grounding, some ideas for a deadline at the end of Week 342, and - most importantly - some homework before a workshop in week 343. That homework lead to some planning, research, and an afternoon of kicking tyres on the internet - amongst other things. I actually need to check what I can say about Longridge - I think I can be reasonably public - so I might describe that more next week.
I spent a day brushing up on the state of web development in 2019. I’m a lapsed front-end developer back from the days when that primarily involved deep knowledge of HTML and awful, awful browsers. (I definitely have built some IE6-compatible sites in my time). Since then, I’ve seen front-end change a lot, and done a nice pile of work on Captionhub with the HTML5 media extensions and spec. But there’s still new things to learn, so I spent an afternoon bringing myself up-to-speed with CSS Grid. Gosh, it’s good, isn’t it? It solves a problem elegantly, and still gives - in many situations - completely acceptable fallbacks when it’s not available. A really elegant API, and a nice bit of technology. I have a small backburner project that I’m using to learn new things on, and I spent some more time on that this week, too; it’s nearly hitting a nice alpha point, so perhaps a concerted day or two in the future will push that over the hill.
It’s nice catching up with technology once it’s a little established. I subscribed hugely to the notion of choosing boring technology. I’ve never been let down by boring technology, and, some days, it’s been reassuring to have my bacon saved by not picking something that goes out of fashion as fast as it came in, or doesn’t have support, or just ended up being the wrong horse. It’s why I still am entirely comfortable shipping Rails projects: it works, it’s expressive for developers, Performant Enough, and gets web-apps based around shipping content to/from databases over HTTP out the door quickly. Uncontroversial. So rather than hurtling to stay up-to-date with trends, I’m comfortable keeping one eye on them, and the other on the Unexciting Present. I’ll read, think, compare, but committing to using them is a very different process. Now that I’m in a lull, it’s time to catch up a little and explore.
I also shipped a few small pieces of code related to things I’ve been doing recently.
wxr_to_jsonis a small command-line node tool for converting Wordpress eXtended Rss to JSON files, simply for ease of processing. It’s a little opinionated - it flattens some one-item arrays into objects - but it works very well for large
WXRdumps; v8 and xml2js are very, very quick.
Secondly, some Ruby. I packaged up a gist by Stefan Daschek into a gem to use as a Capistrano plugin. It’s ideal for deploying static sites (including build processes) with
cap: it makes a clean local checkout, gives you hooks to run build processes, and then uses rsync to move built content to a server (and lets you use all of cap’s versioned deployment tools). The code worked, but it felt cleaner to turn it into a gem, rather than a
libfile floating around my repository. So I finished the bundling job and wrote a pile of documnetation. No idea if either of these will be useful to other people, but they’re easily shared. So let’s do that, then, and perhaps someone else will find them useful. (Why are you using a server and not a CDN for a stic site, you might ask? To avoid yakshaving, primarily. Change one thing at once!)
And that was a week.
12 July 2019
A good week! Most visibly, I updated the case studies on this site with lots of new projects over the past three-and-a-half years. Really pleased to have these write-ups done: they illustrate lots of nice angles on my work, and I’m glad to show off Captionhub in detail.
Under the hood, I’ve moved everything over to Hugo and a lot of static files. That’s been a largely delightful process.
I got a little blocked at deployment - I was hoping to move to some kind of CDN-backed deployment, but things were getting a little complex, so I simplified the problem, and just deployed to my existing host.
That involved some neat wrangling of Capistrano. I like
capsimply because I use it everywhere, be it for Wordpress (with the composer plugin), Ruby, or static sites. It versions directories, allows for rollbacks, and is a neat layer of glue around ssh.
For this site, I took this sample rsync plugin, and then wrote my own cap tasks so that each deploy checks out the clean site to a temporary folder, runs
hugoto build the site, rsyncs that to a server, and finally updates a symlink. It didn’t take long to have that running in my current setup - nice! I might wrap up that rsync plugin into a proper gem next week as nobody seems to have done that yet.
Anyhow, enough about infrastructure.
In new work news, it looks like there’s a writing project - to be known for now as Longridge kicking off next week, and that’ll run in the background for a couple of months. I’ll have more to say on that one in due course, but for now, I scheduled in an early workshop and some onboarding calls.
However, I’m also looking for new projects to run in parallel with that. Ideally, something more technical - prototyping, exploration, or communication and technical consulting. Themes I’m particularly interested in: sound, video, and interaction with those; connected objects and whatever we’re calling “IoT” now; and, perhaps most vitally, tools to empower, enable, and enrich. Work on tools like CaptionHub and instruments like Twinklr and 16n scratches the same itch: giving someone the tool to do work with, to create, and to do things I couldn’t even imagine. I’m interested in continuing to explore that space. get in touch if that sounds like it’s up your street.
11 July 2019
The case studies on this website were getting a little stale. No more! I’ve just published lots of new case studies of individual projects over the past three years.
The big headline that I’m most keen to talk about is a long, detailed writeup of my work on CaptionHub - a project I worked on for 3.5 years, known in this feed as Selworthy. CaptionHub is an online tool for collaboratively captioning and subtitling video. I served as technical lead and pathfinder, taking the initial idea - the “what if?” - to a prototype and beyond into a shipping product, whilst the team grew and the product acquired clients. The write-up is detailed not just because of the length time I worked on the project, but because of the way the product changed as it developed and grew in scale. It’s a project that shows the breadth of my capabilities well, and the finished product is something I’m very proud of.
But there’s lots more in there too. Highlights include: an open source tool for musicians; teaching on the Hyper Island MA; building a digital musicbox; creating a Twitter bot for an installation at the Wellcome Collection.
The write-ups all include extended thought on process, and, of course, link back to the relevant weeknotes that I wrote during the process.
I’m currently looking for new projects to work on: technical leadership, early stage exploration, communication of ideas, are all areas I’m keen to continue in. Topic areas I’m particularly interested in include building tools for creatives and professionals, the bridge between the physical and digital, and audio and video. I’ve written more about my capabilities here.
8 July 2019
I wrapped up the order for Thonk at the beginning of the week when the final parts arrived from China - another excellent job from AllPCB, after some wrangling over my slightly unusual Gerber files. So that was good to get shipped and invoiced.
I also wrapped migrating the Foxfield site to Hugo and Netlify. Whilst I’ve moved a few sites to Netlify now, this was the first Hugo port, so getting things like the RSS feed behaving, and automatic build happening, were good to wrangle on a smaller project than this site.
Speaking of this site, I finally started the big write-up of Selworthy. Unsurprisingly, this took longer than planned - primarily, to find a through-line and plot for describing it, and also to fact-check it. Fortunately, that’s an area where Weeknotes come in handy! I’m going to continue a second pass in week 340, but for now, these are pretty much there.
I had a few phone calls with people I’d been speaking to about a couple of projects. No work really there - so making sure those don’t overrun - but good to consult at early stages if only for half an hour, and perhaps a project may emerge in the nearer term there.
I also spent some time writing some Go. I know, deep down, very, very little Go, but it felt like it might be useful for a script I had an idea for: fast, built-in HTTP library, compiles to a single binary. So I spent half a day writing a script I could probably have written in a hour in Ruby in something new. I greatly enjoyed the process, in the end: the tooling available makes it nicely straightforward, there are good docs, and there was lots to enjoy, such as gofmt everywhere, and slices. A pleasant afternoon feeling competent and productive at code.
1 July 2019
I’m making progress on the latest Thonk order, but kitting is taking a little longer than planned, broken up over the course of days, as parts arrive, or inaccuracies in BOMs reveal themselves (notably - when multiple suppliers are involved). So I push forward, doing what I can when I can, and spending the rest of the time on other work.
That other work has included overhauling some of my websites - and where, and how they’re hosted. I’m trying to work out I can offload to simpler hosting setups in order to reduce my workload and responsibilities. That’s included porting a few sites from Wordpress to Hugo and moving to CDN-style hosting, through services such as Netlify.
Of course, by moving to flat files, I’m not locked in to those services: HTML is HTML, and can be hosted almost anywhere. So the work to simplify and strip down actually means I can be confident that should my hosting needs change, hosting them anywhere else is also straightforward.
Hugo’s been really satisfying to work with. I’ve poked at a fair few static site generators in my explorations before settling on this. The decision making came down to a few points:
- I’d love incremental-build (where only dirty changes are compiled) to be working, but even SSGs that say they support it don’t really. So, rather than prioritising incremental build, why not prioritise pure speed? Hugo is very, very fast.
- I like that it’s a compiled Go binary; it works “everywhere”, is easily portable, and doesn’t require huge module dependencies just to run.
- Because it’s precompiled, it doesn’t have a dynamic plugin structure. That may sound limiting. But, in fact, it forces me to do more with ‘just template language’. The Go template language is a bit idiosyncratic, but I like templating languages a lot, and it forces me to think about markup and structure, rather than just bodging everything with dynamic code. I mean, I used to work with Velocity a lot, so I’m used to getting a lot out of limited tools. (I should note - Velocity was… 13 years ago. I’d rather not ever use it again now…)
And, of course, by getting me to reduce content to Markdown and templates, not only is the output easily portable, but the source code is also relatively easily portable - if I ever move away from Hugo, obviously templates will need rebuilding, but the content is in a neat portable format - not tied up inside a database schema that will need exporting.
So that’s been productive, and largely changed my opinion on SSGs for my own personal use.
I greatly enjoyed playing with Ableton’s Learning Synths earlier in the week. Not just because I’m interested in the subject area, either! It made lots of smart pedagogical choices that I really enjoyed, as someone thinking about explanation and education a fair bit:
- before introducing anything technical, start with abstract tools and aesthetic output. In Ableton’s case, that meant: boxes that make noises as you drag around. Things that sound satisfying. Descriptions of things you’ve heard. Supply the learner with context for what you’re about to present.
- only then is it time to start mapping those things to terminology such as pitch or timbre.
- using pitch to illustrate modulation options - envelopes and LFOs - makes it very easy to hear their input. Even though other controls are more common destinations for modulation, starting with pitch makes understanding the metaphor clearer, quicker.
- making all the later examples Just Work with any attached keyboard attached is a superb idea
There’s so much care in the implementation of the project, too, from the delightful animations through to the richness of the tools. And, a common strand with Ableton: note their willingness to promote applicable knowledge, rather than their own tools, in their education work.
Full marks, and ideas I’m sure I’ll keep thinking about.
And that was Week 338.
24 June 2019
On Wednesday, I went over to Method. They regularly get outsiders in to give lunchtime talks, and they asked me to deliver something for them. I worked up an edit of “How Computers (Don’t) Think”, a lecture I do for my Hyper Island students around “AI”, Machine Learning, and implementations of it such as computer vision or speech recognition. Of course, as the title suggests, it also dives into the language and manner in which these technologies are communicated, unpacking what Marvin Minsky called suitcase words, to understand that more often than not, we’re talking about counting, arithmetic, or statistics rather than new forms of cognition. (Mike Mallazzo’s recent The BS-Industrial Complex Of Phony AI is a good point of reference on this). On the way, we took in MENACE, Clever Hans, and looked at other futures for AI beyond lazy Skynet metaphors.
Re-editing and prepping that took a moderate amount of time, but the effort was definitely worth it: the talk came out the tightest it’s ever been, and better suited to a design-agency audience. I was particularly pleased to be able to talk to one of Method’s machine-learning experts (of the very much non-bullshit kind) who clarified a few points for me but otherwise was highly enthusiastic about the accuracy and clarity of the talk.
I spent an afternoon ordering parts and components for a new Foxfield kits run, which also involved a while fettling gerbers and liasing with my Chinese producer; I do some funky things with silkscreening that aren’t always straightforward to produce. That’s all ordered now, so should be coming together in the coming weeks.
I finished my studio move: all the cruft is now tidied away on storage shelves, and my storage boxes have gone. Glad to have all that dealt with.
Beyond that, some good conversations; notably, a pleasant catch-up with Max from After The Flood, about design, strategy, and the various roles that agencies serve for their clients.