lca2020 ReWatch 2020-02-02

As I was an organiser of the conference this year, I didn’t get to see many talks, fortunately many of the talks were recorded, so i get to watch the conference well after the fact.

Conference Opening

That white balance on the lectern slides is indeed bad, I really should get around to adding this as a suggestion on the logos documentation. (With some help, I put up all the lectern covers, it was therapeutic and rush free).

I actually think there was a lot of information in this introduction. Perhaps too much?

OpenZFS and Linux

A nice update on where zfs is these days.

Dev/Ops relationships, status: It’s Complicated

A bit of  a war story about production systems, leading to a moment of empathy.

Samba 2020: Why are we still in the 1980s for authentication?

There are a lot of old security standards that are showing there age, there are a lot of modern security standards, but which to choose?

Tyranny of the Clock

A very interesting problem solving adventure, with a few nuggets of interesting information about tools and techniques.

Configuration Is (riskier than?) Code

Because configuration files are parsed by a program, and the program changes how it runs depending on the contents of that configuration file, every program that parses configuration files is basically an interpreter, and thus every configuration file is basically a program. So, configuation is code, and we should be treating configuration like we do code, e.g. revision control, commenting, testing, review.

Easy Geo-Redundant Handover + Failover with MARS + systemd

Using a local process organiser to handle a cluster, interesting, not something I’d really promote. Not the best video cutting in this video, lots of time with the speaker pointing to his slides offscreen.

 

Restricted Sleep Regime

Since moving down to Melbourne my poor sleep has started up again. It’s really hard to say what the main factor driving this is. My doctor down here has put me onto a drug free way of trying to improve my sleep, and I think I kind of like it, while it’s no silver bullet, it is something I can go back to if I’m having trouble with my sleep, without having to get a prescription.

The basic idea is to maximise sleep efficiency. If you’re only getting n hours sleep a night, only spend n hours  a night in bed. This forces you to stay up and go to bed rather late for a few nights. Hopefully, being tired will help you sleep through the night in one large segment. Once you’ve successfully slept through the night a few times, relax your bed time by say fifteen minutes, and get used to that. Slowly over time, you increase the amount of sleep you’re getting, while keeping your efficiency high.

Moving to Melbourne

Now that the paperwork has finally all been dealt with, I can announce that I’ll be moving down to Melbourne to take up a position with the Australian Synchrotron, basically a super duper x-ray machine used for research of all types. My official position is a >in< Senior Scientific Software Engineer <out> I’ll be moving down to Melbourne shortly, staying with friends (you remember that offer you made, months ago?) until I find a rental near Monash Uni, Clayton.

I will be leaving behind Humbug, the computer group that basically opened up my entire career, and The Edge, SLQ, my home-away-from-home study. I do hope to be able to find replacements for these down south.

I’m looking at having a small farewell nearby soon.

A shout out to Netbox Blue for supplying all my packing boxes. Allll of them.

Actively looking for work

I am now actively looking for work, ideally something with Unix/C/Python in the research/open source/not-for-proft space. My long out of date resume has been updated.

Access and Memory: Open GLAM and Open Source

Over the years of my involvement with library projects, like Coder Dojo, programming workshops and such, I’ve struggled to nail down the intersection between libraries and open source. At this years linux.conf.au in Sydney (my seventeenth!) I’m helping to put together a miniconf to answer this question: Open GLAM. If you do work in the intersection of galleries, libraries, archives, musuems and open source, we’d love to hear from you.

South Coast Track Report

Please note this is a work in progress

I had previously stated my intention to walk the South Coast Track. I have now completed this walk and now want a space where I can collect all my thoughts.

Photos: Google Photos album

The sections I’m referring to here come straight from the guide book. Due to the walking weather and tides all being in our favour, we managed to do the walk in six days. We flew in late on the first day and did not finish section one of the walk, on the second day we finished section one and then completed section two and three. On day three it was just the Ironbound range. On day four it was just section five. Day five we completed section six and the tiny section seven. Day six was section eight and day seven was cockle creak (TODO something’s not adding up here)

The hardest day, not surprisingly, was day three where we tackled the Ironbound range, 900m up, then down. The surprising bit was how easy the ascent was and how god damn hard the descent was. The guide book says there are three rest camps on the descent, with one just below the peak, a perfect spot for lunch. Either this camp is hidden (e.g. you have to look behind you) or it’s overgrown, as we all missed it. This meant we ended up skipping lunch and were slipping down the wed, muddy awful descent side for hours. When we came across the mid rest camp stop, because we’d been walking for so long, everyone assumed we were at the lower camp stop and that we were therefore only an hour or so away from camp. Another three hours later or so we actually came across the lower camp site, and the by that time all sense of proportion was lost and I was starting to get worried that somehow we’d gotten lost and were not on the right trail and that we’d run out of light. In the end I got into camp about an hour before sundown (approx eight) and B&R got in about half an hour before sundown. I was utterly exhausted, got some water, pitched the tent, collapsed in it and fell asleep. Woke up close to midnight, realised I hadn’t had any lunch or dinner, still wasn’t actually feeling hungry. I forced myself to eat a hot meal, then collapsed in bed again.

TODO: very easy to follow trail.
TODO: just about everything worked.
TODO: spork
TODO: solar panel
TODO: not eating properly
TODO: needing more warmth

I could not have asked for better walking companions, Richard and Bec.

In Memory of Gary Curtis

This week we learnt of the sad passing of a long term regular attendee of Humbug, Gary Curtis. Gary was often early, and nearly always the last to leave.

One  of Gary’s prized possessions was his car, more specifically his LINUX number plate. Gary was very happy to be our official airport-conference shuttle for linux.conf.au keynote speakers in 2011 with this number plate.

Gary always had very strong opinions about how Humbug and our Humbug organised conferences should be run, but rarely took to running the events himself. It became a perennial joke at Humbug AGMs that we would always nominate Gary for positions, and he would always decline. Eventually we worked out that Humbug was one of the few times Gary wasn’t in charge of a group, and that was relaxing for him.

A topic that Gary always came back to was genealogy, especially the phone app he was working on.

A peculiar quirk of Humbug meetings is that they run on Saturday nights, and thus we often have meetings at the same time as Australian elections. Gary was always keen to keep up with the election on the night, often with interesting insights.

My most personal memory of Gary was our road trip after OSDC New Zealand, we did something like three days of driving around in a rental car, staying at hotels along the way. Gary’s driving did little to impress me, but he was certainly enjoying himself.

Gary will be missed.

 

Software Carpentry

Today I taught my first Software Carpentry talk, specifically the Intro to Shell. By most accounts it went well.

After going through the course today I think I’ve spotted two issues that I’ll try to fix upstream.

Firstly, command substitution is a concept that is covered, and used incorrectly IMO. Command substitution is fine when you know you’re only going to get back one value, e.g. running an identify on an image to get its dimensions. But when you’re getting back an arbitrary long list of files, you’re only option is to use xargs. Using xargs also means that we can drop another concept to teach.

The other thing that Isn’t covered, but I think should be, is reverse isearch of the history buffer, it’s something that I use in my day to day use of the shell, not quite as much as tab completion, but it’s certainly up there.

A third, minor issue that I need to check, but I don’t think brace expansion was shown in the loop example. I think this should be added, as the example I ended up using showed looping over strings, numbers and file globs, which is everything you ever really end up using.

Software Carpentry uses different coloured sticky notes attached to learners laptops to indicate how they’re going. It’s really useful as a presenter out the front, if there’s a sea of green you’re good to go, if there are a few reds with helpers you’re probably OK to continue, but if there’s too many reds, it’s time to stop and fix the problem. At the end of the session we ask people to give feedback, here for posterity:

Red (bad):

• Course really should be called Intro to Unix rather than bash
• use of microphone might be good (difficult to hear, especially when helpers answer questions around)
• Could have provided an intro into why  unix is advantageous over other programs
• grep(?) got a bit complicated, could have explained more
• start session with overview to set context eg. a graphic
• why does unix shell suck so much, I blame you personally

Orange(not so bad):

• maybe use the example data a bit more

Green(good):

• patient, very knowledgeable
• really knew his stuff
• information generally easy to follow. good pacing overall good
• good. referred to help files, real world this as go to for finding stuff out (mistranscribed i’m sure)
•  good pace, good basis knowledge is taught

 

 

South Coast Track

I’m planning on doing this walk before linux.conf.au 2017. I’m interested in a couple of experienced hikers to join me. It starts at Melaleuca and finishes at Cockle creek.

Important Points

This is a reasonably serious hike, you need to have plenty of multi day hike experience to join me.

• Estimated dates: 8th Jan to 15th Jan 2017
• Eighty five kilometre hike done over seven days.
• Tides and wet weather may add a day or three.
• Fly in to the starting point (yeah, it’s remote)
• Carrying own food and sterilize own water.
• Everything you take in, you take out.
• Taking care of your own poop along the way (think shovel).
• Camping overnight, no cabins
• Bus back to Hobart (three hours)

Todo:

• Organise Park pass.
• Organise flights and fuel.
• Organise bus.
• Try not to scare everyone from joining me.
• Order guide book. (done)
• Beg/borrow/steal/hire epirb (sourced, need to ship)

Links:

•  https://www.southcoasttrack.com.au/
• http://www.parks.tas.gov.au/?base=2265

YOW Brisbane 2015

The YOW conference is very kind to local meetup organisers, I was lucky enough to be offered a ticket in return for introducing a couple of sessions.

Monday

Keynote: Adrian Cockcroft

Complexity, understanding, composition and abstraction.

Past, Present and Future of Java: Georges Saab

Some of the new fp/multi core stuff slowly coming down the pipeline. I’ve always had high expectations for Java and the surrounding environment, but every time I’ve used it I’ve been very disappointed. There’s a lot to be said for backwards compatibility, but not at the cost of destroying all the good will your development community has. The changes portrayed in this talk are quite interesting.

Play in C#: Mads Torgersen

This was a highlight of the conference for me. The Roslyn project basically inverted the Microsoft compiler from a sink to a filter which lets it be hooked up directly to the IDE. The live example was adding a linter to the IDE to complain about blocks of code not in brace extensions, complete with one click fixup. It was all very impressive.

Writing a writer: Richard P. Gabriel

Generating poems that get judged to be written by humans, all in lisp of course.

Keynote: Don Reinertsen

This was a very interesting discussion on the natural reaction in an uncertain world: making systems robust. At the very best, the most robust system (robustest? :) will be able to handle the most chaotic world and bring system performance back to normal. This talk asks us to think about the notion of a system that can actually improve in a chaotic world. The theoretic model is based on the financial idea of increasing risk implying increasing returns.

The Future of Software Engineering: Glenn Vanderburg

This was a very interesting talk on the nature of engineering, and how software engineering fits into the discipline. A highlight.

The Miracle of Generators: Bodil Stokke

This was an FP talk, I’m not a fan of bait-and-switch talks.

Tuesday:

NASA Keynote: Anita Sengupta and Kamal Oudrhiri

It’s interesting to be in a room full of engineers being exposed to different engineering requirements.

Agile is Dead: Dave Thomas

A great simplification of the underlying ideas of how to have agility.

Sometimes the Questions are Complicated, but the Answers are Simple: Indu Alagarsamy

A highlight of the conference overall, a talk about a healthy family culture butting up against backwards societal culture.

Keynote: Kathleen Fisher

Formal processes work, but we’re decades off being able to use them for day to day work.

Always Keep a Benchmark in your Back Pocket: Simon Garland

Some rules to keep in mind around designing  benchmark, plus the idea of always doing benchmarking as a way of defending development work to management keen on outsourcing.

Transcript: Jonathan Edwards

One of the talks I chaired.  A very interesting document and form based programming language for non-programmers to use, in the style of hypercard.

The Mother of all Programming Languages Demos: Sean McDirmid

One of the talks I chaired. More interesting ideas coming out of Microsoft. This was heavily based on physical interfaces, I struggled to think how it would apply to regular programming.