“When the entire household goes to bed, I do curl development for 2 more hours.”

"Getting off the ground" is for me, a great big open-ended architectural decision making process with an ocean of second-guessing myself. It's very very likely to keep me awake for hours reevaluating everything I did in that last couple of hours.

Smaller pieces of work are different. Actually fixing a bug has a fairly well defined "ending" and I can sleep easy basking in the warm glow of satisfaction. Even getting part way through implementing a feature is a thing that doesn't put my brain into "wide open exploration of all possibilities" mode. It's maybe not so easy to bask in achievement and sleep as a quick bug fix, but at least the "code dancing in my head" is a fairly narrow path of implementation I've probably already got the structural problems solved for.

For me - starting projects need empty weekends.

I've found that giving myself a solid hour to wind down helps immensely with slowing my brain down and letting me sleep more peacefully. Sometimes I read, or listen to music, or watch youtube, or have a beer or an edible and just hang with out, or just tidy the house or something. Whatever it is, it's not work, it's not in my office, and it's not hard. I've found I've been sleeping far better because of it.

Even on nights that I work super late on some head scratcher, with a full 12-15 hours at my desk, I still give myself the hour before going to bed.

This will invariably affect my sleep with either no sleep at all for 1-2 hours or dreams of fixing an issue over and over again resulting in me being less rested the next day.

I'd love for there to be a quick way to 'snap' out of programming mode but after over a decade of programming I haven't really found one (except for having a couple of pints...).

I m in a phase where the code at work interests me so much that I think I sleep 5 hours a day, the rest I touch that code. Non stop, and I have a kid.

Thank god it s usually just a phase :D

Modern life is just so depressing, we have so little free time. Everything must be scheduled, compartmentalized, etc

So, 6.5 hours of sleep though? I need either 7 or 9, but 8 does not work for me.

So, I would be looking at 5a.

Let me ask you, do you immediately start coding, or do you need a few minutes to ‘warm up’?

What is the morning ritual, if you don’t mind me asking?

But the screwed up sleep cycles from years and years of this are starting to take their toll, so I've been trying everything I can to replicate that peace of mind during daylight hours. No silver bullet yet.

There's less local news. There's nobody to call or text. There are no deliveries. Slack is quiet. My email is quiet. Our repos are quiet. The dev logs are quiet. My family is asleep so I know they're good. If I need a minute to think, I can sit on my front porch and not worry about being distracted.

Now that I have a young kid, it's really hard (and generally selfish) to work those hours. I miss it.

If people are up, I can’t lock in; there’s a constant distraction somewhere in my mind.

If I’m the only person awake in the house, I seem to hit flow state with ease.

Personally, I find I sometimes have trouble accepting good enough solutions. I want more structure, a better defined start. At night I can drop into a productive flow state with much less resistance, ignore many of the day-time wants. The hours run short; let's seize what we can from the waking hours. As well that "inspiration," I think there is probably a physiological barrier lowering that happens as we go into the night. As we recede from the gods' light, as it were.

I could see creating quality implementations of each of those taking that long.

https://curl.se/

> we add features, improvements and bug-fixes continuously. With 240 command line options , 26 transfer protocols and countless combinations of things done over a crazy Internet that's just the nature of it all...

Curl has 170k lines of code, so it's bound to require some maintenance.

For me, 7 kids was the point at which I could no longer effectively maintain procps.

IMO your comment is even more inspiring than the thread itself :)

I’m guessing most our our personal metrics for what makes software “good” are only somewhat aligned with capital.

I'd imagine manual laborers sleep much more soundly, and are less likely to code in their down time. (but maybe!)

[0] https://medium.com/@hfuecks/so-we-went-with-a-hamburger-menu...

I mean, sure, you can do Project Euler (or many other things), but you can also work on cool projects. I'm writing a superoptimizer (and I have C++ coding stuff at work). For both, I make a conscious effort to try to figure out better ways of doing that stuff (rather than just banging out ancient idiomatic C++ that isn't modern or good) - it's easy to slip into "doing things a dumb but quick way" that doesn't teach you much about the language or coding.

I have a similar attitude to push-ups, honestly. I'd rather bench press once a week and do a sport than be constantly banging out push-ups. In both things, you quickly get to the point where you're hitting diminishing returns from the "push-up" type of operation, unless your goal is "to be the best at exercising".

If we're comparing someone who couldn't complete 10 pushups or do a basic Project Euler problem with someone who can, I'd say "sure, that first person needs pushups (or programming equivalent)". I'm not sure once you're well beyond that point whether it makes sense to do that many more "pushups", unless you are going to a push-up competition in a month's time.

What are you trying to accomplish with C++?

If you’re looking to have a better command of syntax, then do advent of code or something aimed at that. If you’re trying to implement algorithms in C++, get an algorithms book and implement them in C++. If you’re trying to get better at designing and architecting C++ code, look at good implementations (easier said than to find).

What ever you practice, that’s what will improve, so it’s best to align that with your goals and work backwards.

Athletes tend to have different phases of their training. Some periods of pre-season training, season, post-season where there may be a greater emphasis on things like strength, endurance, power.

I think building is good, but there must be some other phases too. You can’t just build all year or else you might get stuck in a local maxima. So, there should be phases of learning, building, refining. Again, aligned with what your end goal is.

I don't think performing a specific exercise is the route here. It's probably best to work backwards and find projects/challenges you find interesting and implement them in c++. You'll inevitably build new muscles + strengthen existing ones and most critically, be motivated to do it.

Once you're not a beginner, programming mostly becomes about the specific domain youre working on, rather than the code. So improvement comes from a better understanding of banking or file systems or whatever you're doing, rather than cross-training in some proxy thing the way you do with exercise.

https://projecteuler.net

Is it really not cool to wish that developers would get compensated for their time if their work is valuable?

I understand that the developer chose the license and distribution model, but if they hadn't chosen this model, then the community would have settled on whatever the best solution was that was permissive enough that they didn't have to pay for it. Passion projects in OS are a race to the bottom compensation-wise, due to dynamics outside of the control of the developers building them.

Usually, this is due to decisions the author made, and the author is in full control of changing the situation.

I come from a society where it's normal for people to give stuff with no strings attached, and then later claim I have a moral obligation to pay back for all the benefits I received. It sucked being in that society.