jlj

The musings of an aspiring carver of space

Deep Work

I'm just starting the subject book, by Cal Newport. I've been thinking about reading Newport for quite some time: CGP Grey recommended So Good They Can't Ignore You on a very early episode of HI, as I recall. Rob Muhlestein (aka rwxrob) recommending Deep Work as part of his RWX.GG knowledge project was the nudge I needed.

I begin this endeavour with more than a bit of trepidation.

Am I cut out for deep work? I honestly don't know. Back in the days of pencil and paper, I remember accomplishing most things through a facility for memorisation, and the adrenaline surge of a time-crunch. (I was an astonishing procrastinator.) I suppose, thinking about it, I did produce some Dungeons & Dragons campaigns of staggering detail, requiring many hours of concentrated work. But that isn't the same thing as pushing my brain with a heavy cognitive load for an extended period of time.

But I must try. As I've said before, my training programme is not letting up, and my current, chaotic approach is reaching the end of its utility.


I finished Notes On A Nervous Planet by Matt Haig last night. I think I mentioned that a friend had recommended it, knowing about my problems with anxiety. While I think there's a high likelihood — on any particular random opening of the book — of your eyes falling on something rather facile, I would be lying if I said that it hadn't helped. My zettelkasten, only started in earnest when I was already two-thirds of the way through the book, still has notes on half a dozen pages of Haig's thoughts.

I'm still anxious, though. The degree varies from day to day, spiking when I think about leaving the house (which I rarely do). Another source is looking at what's required of me professionally over the coming months. I'm hoping Newport will help me get a handle on that.

End of Day 42

jlj@Fosstodon #100DaysToOffload

I'm writing this as part of the 100 Days To Offload project; join us at: https://100daystooffload.com/

Happy Father's Day!

This one will be brief.

I'm a lucky man, and I've spent the day feeling very grateful for that, surrounded by my loved ones.

I've had a complicated relationship with my own father, but even in this, I'm luckier than many: he's survived to my middle age, where we've both come to realise that we have pretty much diametrically opposite views of our shared past, and that there's little to be gained by dwelling on that now.

This, in turn, has galvanised my commitment to my own children. I will make mistakes; likely many of them. But my children will know that I regret each and every one of them, and still love them with all my heart. They will know this because I will tell them. Often. Probably too often.

They will know how special they are to me. And that however badly this world and the people in it hurt them, betray them, disappoint them, I will be there to listen. And to share in what I hope will be their much-more-prevalent joys.

End of Day 41

jlj@Fosstodon #100DaysToOffload

I'm writing this as part of the 100 Days To Offload project; join us at: https://100daystooffload.com/

Derek Sivers is nice

As usual, the weekend (so far) has been filled with many lovely things that I won't be blogging about. And then we decided to finish Picard tonight as well. (I do have some thoughts on it, of course, but none worthy of the spoiler warning that would have to go with them. Drop me a line if you're interested in this diehard TNGer's take.)

I ran out of space with my rclone project: it's a hard 15GB limit that doesn't seem to have touched the estimates on my actual Google account. I can't imagine a service account I create gets its own Google Drive allocation, but I don't know how else to explain what is going on. Suffice to say that it's been fun, but I'll have to think about what I might want to use that space for at another time. For now, I've picked up a 128GB SanDisk thumb drive for £15 and copied the Timeshift snapshot to it. (It's extraordinary, to me, that it takes a third of the physical space that the 4GB thumb drive that I still use does.)

I now feel like I'm on fairly firm ground with my backups. I'll still need to test them, down the road, if I can be bothered. To be honest, as long as I have a copy of my photos, videos and documents, rebuilding just seems like a chance to practice system administration. (Some would argue that creating and restoring functioning backups is a very important skill in that suite, of course. One that requires some effort to practice.)


I started watching Killswitch (2014) last night, curious to see more of Aaron Swartz. I'll write about that once I've finished it. I will say that I've already ordered a copy of Tim Wu's The Attention Merchants off the back of what I've seen, though.

I'll close with a nod to my Now page and Derek Sivers. You may have noticed that the former is linked at the top of my site; it's a .plan file for the 21st century. And I mean that in the best possible way. I'm looking forward to exploring Sivers' body of work... (OK, I've just googled him now, and, my first thought is that I can't believe how accessible and friendly he is over email. Wow. OK, I'll probably have more to say on that in a future post; I certainly have a title for this one now.)

End of Day 40

jlj@Fosstodon #100DaysToOffload

I'm writing this as part of the 100 Days To Offload project; join us at: https://100daystooffload.com/

The Internet's Own Boy: The Story of Aaron Swartz

I watched this for the first time last night.

Somehow, I hadn't even heard of Aaron Swartz.

And, while I let that sink in, it's worse than being a fan of networked technology and not knowing: I was active in the digital rights scene back in Ottawa at the time too. I wasn't just reading Geist: we were meeting, as part of the group trying to get Bill C-61 amended. And we were all reading Lessig, Doctorow, and were intensely interested in Eldred v. Ashcroft.

I loved Creative Commons. It was all over my creative writing site at the time. I had no idea Swartz wrote that code.

I've tried to let myself off the hook a bit: maybe all these people he worked with — on RSS as well, and even Markdown, apparently, although that's from Wikipedia — were trying to keep him from the limelight because he was still a minor. I don't know.

It's a bit easier to understand how I missed his death: my personal life was a shambles in late 2012/early 2013. I was barely keeping it together, even with my head down, just trying to get by.

The film is as interesting for who it wasn't able to get, I think. I found the whole topic to be pretty upsetting, so I won't be doing any research on this or other points. For me, that they got his parents talking was enough: and that made it all the more heartbreaking, of course.

He was so eloquent! Even on live television. That, to me, speaks of his brilliance more than anything else: that he was able to answer questions so well, on the spot. Likely what I consider a home-run answer would've had him shaking his head afterwards, wondering why he hadn't said this or that. I bet his shower thoughts were world-altering.

And that's the saddest part of all: Clay Shirky referenced it in one of the few articles I read about Swartz's death: everyone will miss out, is currently missing out, on all the other things he would've done. He was just getting started. It's awful; just awful.

I'll close with Sir Tim Berners-Lee's beautiful epitaph:

Aaron dead. World wanderers, we have lost a wise elder. Hackers for right, we are one down. Parents all, we have lost a child. Let us weep.

RIP Aaron Swartz 1986 – 2013

End of Day 39

jlj@Fosstodon #100DaysToOffload

I'm writing this as part of the 100 Days To Offload project; join us at: https://100daystooffload.com/

Is Python as esoteric as it seems?

First, the new colours: Awful? Nice? Strange? If I'm risking the little traffic I'm getting now with this, I'd appreciate a quick heads-up/@mention. ;–) I was playing around with Khroma late last night, and I'm still sleepy now, so I'm willing to accept that my judgment may be flawed. [Edit: I'm going back to the drawing board.]


My zettelkasten project is off the ground: I've been using Obsidian for a few days now. Just for note-taking — modelled roughly on the workflow outlined in Effective Remote Work's video on the topic — because working with new software while also learning the nuances of a more involved workflow seemed audacious at this stage. If there's one thing I've learned from blogging over the years, it's just bash it out: get something down, and keep that ball rolling; you can edit — heavily, if necessary — later.

One stumbling block to that process has already surfaced: I thought I'd blogged about all this, because the interface and thought process I use for both are, well, identical. But, no, a few keyword searches later revealed nothing on the blog; this is another process I'll need to refine as my slip-box grows.


That rclone process I talked about yesterday is still running, by the way; some 32 hours later. This is both surprising and fine: I'm really hoping that most recovery scenarios will involve using local snapshots; this is yet one more nightmarish aspect of the nightmare scenario of a house-fire or something, I guess.


I've talked about how I'm thinking about learning Python. I'm questioning the wisdom of that lately: every time I try to play with some code, or even get someone else's up and running, I end up on Stack Overflow nine or ten times an hour. My thought process, in no particular order, goes:

  • Should I set up a virtual environment?
    • I'll run pip when I want pip3 then; better do it.
  • Have I activated it yet?
  • How should I install this? Easy install?install setup.py .? With a whl file instead?
    • Is using pip ... setuptools --upgrade associated with the former? Is upgrading wheel all I need to do to get the latter working?
  • Is this pyyaml error related to the pip install, or to a missing dependency in my wider Linux installation?
  • What's Anaconda?
  • Do I still need to worry about generating a requirements file?

On and on. Worse still, nothing sticks because I'm not working with this stuff enough. And, yes, I'm exaggerating a bit to make a point. Maybe it comes down to what you learned on. When I'm compiling C code, I'm looking to the developer to point out dependencies; I'm happy pulling down libraries based on their documentation, or on subsequent build errors, even linking libraries symbolically as required. That may make as little sense to someone else as the various building blocks of any Python environment do to me, I suppose.

End of Day 38

jlj@Fosstodon #100DaysToOffload

I'm writing this as part of the 100 Days To Offload project; join us at: https://100daystooffload.com/

Remembrance, versus reverence and normalisation

Following on from the news about terminology changes in the codebase of both GitHub and OpenZFS, I joked in our company's chatroom about how our new hires like me might be the last contingent to test on Splunk's current cluster terminology. To my surprise, a colleague quickly replied, saying that such changes are indeed afoot.

I have to admit, the urge to roll my eyes was strong. In fact, my response was, “Satire is hard these days.” But, on second thought, there is something to be said for these changes, if less about the motives, so publicly expressed, behind them. A few years ago, I wrote about the Confederacy statues being demolished in the American south. To paraphrase, there is a very important difference between remembrance and reverence. We are not cleansing history with these acts, as some have argued, but failing to revere them any longer.

Now, few would call referencing the terms of slavery — in software and code, devoid of any other context — reverence. I could see an argument being made for calling it normalising, though. And, to go further, that the outcomes from both reverence and normalisation are exclusively somewhat to extremely negative.

As I get older, I realise how difficult it is to objectively assess things that have been a part of one's experiences and observations for most of one's life. Whether something came to be five years before your birth, or 50 years before, will be indistinguishable to you. (Well, let me quickly clarify that that is the perspective of a cis white male, very obviously, and, far from coincidentally, likely also the perspective of the majority of eye-rollers right now.) But that fact alone is poor rationale for it to outlive you as well.


I did have some success with the home set-up today too: rclone is currently backing up a Timeshift snapshot to Google Drive using a service account I set up for the purpose. I wonder how long the term clone will be in common use.

End of Day 37

jlj@Fosstodon #100DaysToOffload

I'm writing this as part of the 100 Days To Offload project; join us at: https://100daystooffload.com/

Timmy's ain't been quaint for a long ole time

My son decided that 5am was go-time this morning. My Splunk revision has suffered as result, I think it's fair to say. Still, I feel more comfortable with search head clustering now, so it wasn't a total loss. For completeness, in a disaster recovery scenario, whether the remaining search heads are an odd or even number isn't relevant (as far as I've been able to discern): it's whether 51% of the original membership remains. If so, the election of a new captain will proceed without issue. If not, manual intervention is required; likely in the form of designating a static captain temporarily.

Fascinating, I know.

Don't even get me started on the criteria under which a master node will rebalance the indexer cluster, or what triggers a change in its generation (ID).


But you didn't come here to revise Splunk. (I hope — may God have mercy on your soul otherwise.)

I'm Canadian. I think that's come up before, but I wanted to put it out there again, as Tim Horton's is a national institution to our older generations (of which I am most certainly a member). I mean, it's been an American company for, oh, twenty years now or something, and stopped selling doughnuts with coconut shavings thirty years ago, so, really, why would anyone care? ;–) The fact is, “double double” is (still?) part of our lexicon (meaning coffee with two portions of cream and two portions of sugar, I know you were dying to know).

And Tim Horton's is in the news. Well, in the Financial Post. For avidly tracking the users of its app. I was shocked by the bit on server-side location processing, or the tools at the command of third-parties in the event of you preventing an app from accessing your GPS data. The type and number of events generated around user movements — including confidence factors around why the user might be where they are — is chilling. I mean, this makes loyalty cards mickey mouse; an increase in customer data of, what, tenfold? More?

And so, folks won't actually be rolling up any cup rims in the future, it seems. (Plenty of Canadians' favourite new car dream.)

End of Day 36

jlj@Fosstodon #100DaysToOffload

I'm writing this as part of the 100 Days To Offload project; join us at: https://100daystooffload.com/

Full horizon

Well, my next exam is booked: Splunk Enterprise Certified Architect, in a week's time. Closed book, as per usual, but this one is longer than earlier ones: 90 minutes. The blueprint is pretty intimidating, frankly; especially when you consider that the practical lab test I just finished purposely doesn't include search head clustering.

I need to brush up on so much stuff this week, but the bogeyman in front of me today is search head captain elections; specifically, what happens when one of two sites falls over, taking the current captain with it, and the remaining site has an even number of search heads. I remember the instructor said it's actually best to have a third (logical) site in this scenario with an additional search head, to break the inevitable deadlock in the new election.

But I don't understand why the deadlock occurs in the first place. It just stuck in my head because the instructor said there's a myth circulating that you can only have an odd number of search heads in a SHC; not true, but you do have to be careful. More work to be done.


Then it'll be on to writing more Terraform code for our lab environment. Who said furlough was bench time? As long as it's primarily about my learning, there's no shortage of things to be done.

I need to start planning some exercise in my routine, though. I'm convinced that's why I'm sleeping so poorly. I could put my head on my desk right now and be asleep in five minutes. This is partly because the house is so quiet. Today is my daughter's first full day at nursery. I can't remember the last time I was surrounded by so much silence that I wouldn't break for fear of waking a child.

End of Day 35

jlj@Fosstodon #100DaysToOffload

I'm writing this as part of the 100 Days To Offload project; join us at: https://100daystooffload.com/

Making oneself heard

Another day with little time behind the keyboard: my partner goes back to work tomorrow (remotely), so we made sure to have plenty of family fun in the garden. My daughter — 14 months now — may not have words yet, but she makes her wishes known: from the time she gets up until I'm singing her to sleep, she's asking to be picked up so she can point at everything around her. And if you don't make the noise she expects for, say, that painting of a ship, or, say, starting singing an In The Night Garden theme for the character she's picked out on her lampshade, prepare for an earful.


I did have some success today, however: my RSS feed seems to be displaying properly in Conky now. After a bit of research, it seemed likely that I was running up against Reddit's very low tolerance for bots; it was suggested that distinguishing the call with a user-agent string that included my username might help. For this, I switched back to my bash script — well, it's the one written by Bill Woodward that seems to have made the rounds — and modified it to take an optional user-agent string that it would then add to the curl command using the -A flag.

And I haven't seen “Too many requests” or a blank section since.


dvavasour@fosstodon.org wrote a nice little post on Dasher today. I hadn't heard of it before. I'm really impressed with what I've seen so far. It almost makes me wish I was back in my old job. (Well, not really; that was backbreaking. Maybe if I was working with the Augmentative and Alternative Communication devices.) It seems like a lifetime ago, but I did work for National Star as a facilitator for five years up until the beginning of this one. In a nutshell, you're there to help the students — who often have very complex disabilities — advance in their chosen college programme; everything is individually tailored.

One of the students I worked with a lot used an AAC device mounted to his wheelchair to communicate. Through the use of paddle switches in his headrest, he could select concepts and then words as the device cycled predictably through a grid of icons. Cognitively, he could converse with ease, with anyone. But he needed a lot of lead time, if he had questions he wanted to ask or statements he wanted to make. Constructing the sentences was very time consuming. He had to deal with some athetoid movements, but, usually, once he was comfortable in his chair, these were infrequent. I bet an eye-gaze device running Dasher would speed up his composition process tenfold, even accounting for some sort of error-checking process that would allow him to fix mistakes.

I'm going to see whether my former colleagues are aware of Dasher.

End of Day 34

jlj@Fosstodon #100DaysToOffload

I'm writing this as part of the 100 Days To Offload project; join us at: https://100daystooffload.com/

Debugging, and tech manuals

Didn't spend much time at the keyboard today. Chose to watch Picard this evening instead. (Don't worry; no spoilers.) What time I did spend there was mostly consumed by trying to debug my Conky config; yeah, I spoke too soon yesterday.

It seemed like my RSS script was being called too frequently; I'd grabbed the code from elsewhere, and “Too many requests” kept appearing in the window. I decided to switch back to the native RSS support, but that seemed to consistently return nothing. After playing around with Epsi's Lua code for debugging Conky, I confirmed that it too was eliciting the same response from Reddit (in the form of a 429 status in this case).

That's as far as I got because, while thinking about another RSS feed I might inadvertently abuse as part of this debugging process, I suddenly remembered LWN.net. I started reading it back in the late 90s — right around its inception, it seems, in 1997, although I didn't remember that — and then just stopped at some point, I guess; much like Slashdot. I couldn't believe it was still going, and I spent quite a while reading the archives. It, unlike other concepts on the 'net in the late 90s, has held up very well.

I read that, while they weren't struggling (yet), many of their former subscribers cited losing their jobs as their reason for cutting back. Well, I bought a year's subscription on the spot. That sort of experience, of quality, is effectively irreplaceable. I read a comment on HN the other day about that person's work at O'Reilly in its heyday. I bloody loved those books. Unix in a Nutshell? Advanced Perl Programming? All twenty or so of the X Window System bleedin' encyclopedia? With the exception of the latter — which was always scattered around the office as well as any treasure hunt — these bibles never left my desktop; never even got put back up on my shelves.

And, as much as I love Jeff Atwood, Stack Overflow killed them. Not directly, obviously, but as a consequence of us, collectively, failing to value experience, expertise and, just, hard work. The (wo)man-hours that went into these books — as laid out by the commenter — were staggering; far exceeding the most comprehensive entry I've seen in Stack Overflow. I used to believe that reputation systems would save the world; the cream they curated would be indistinguishable from god-sent.

I've come to my senses.

Simply put, I think there's a lot that collaboration — and the tedious grind of peer review and an editing department — bring to the table that reputation systems can't replicate; or at least the ones I'm familiar with. And that's nothing against them; I've seen some brilliant ones, that do indeed curate top-shelf content. But how did we decide they would replace our reference material, our ground truth in technology? Well, I think the answer is that we didn't; we just failed to step in when it became clear that they would perish without adequate investment.

When did this become a rant? I need to hit the sack.

End of Day 33

jlj@Fosstodon #100DaysToOffload

I'm writing this as part of the 100 Days To Offload project; join us at: https://100daystooffload.com/

Enter your email to subscribe to updates.