How to Drastically Boost Your Developer Productivity | Hacker News Reader

How to Drastically Boost Your Developer Productivity

expose.sh

34 pointsaussieguy123423 comments

This is the exact same silly argument that people at Reddit used to justify buying the new macbook pros for all their developers. It starts off by implying that developers are 100% efficient w/ their use of time and any time spent saved will be spent doing something productive, instead of say: browsing Hacker News. And then it uses that implication to show off exciting statistics like "you will gain 5 weeks of developer time!".

I really feel like this is delusional thinking in the software industry and it seems to be widespread.

I've worked at places with shit pcs for developers, it was maddening. It cost so much time that would easily cover the cost of new machines. Also I dread ever moving to a company that uses VMs in the cloud for developer pcs

I don’t really agree; small, continuous disruptions that happen throughout the day can totally kill any sort of flow state that I otherwise would have gotten in, thereby significantly reducing my productivity. On the other hand, fast edit and refresh cycles really keep me in the zone and keep me productive. When “delusional” thinking is widespread it may be worth taking the time to understand why rather than dismissing it.

Know thy time. A specific and comprehensive understanding of how one’s time is spent is the precondition for choosing and following through on effective behaviour change.

Completely agree. It's an insane idea that developers should try to work as quickly as possible to start with. Why should they?

I automate things because I don't like doing manual tasks (because it's boring). I don't do it to save time for my company.

I think it is more about freeing up your mental energy and concentration than actual time saved

Counting seconds like that isn't exactly correct in my experience. I find that I actually don't need to go that fast on the whole.

For instance, reading HTML docs in a browser with a text editor vs using a fancy IDE. Metrics say IDE wins, but I don't think those seconds shaved are in the critical path. And there are other benefits to the non-IDE workflow.

  > If automating one repetitive task saves you 36
  > seconds and the equivalent of a working week per
  > year, automating 5 similarly time consuming tasks
  > will get you back 5 weeks per year.
  > 
  > How would I solve this specific problem? Use a
  > newer framework with Hot Reload support.

Why did I never think of completely rewriting my old Java monolith using a new framework to save 5.4 days of dev time?! Revolutionary!

Use a real debugger instead of printing variables

"Print real variables instead of using a debugger"

That's a bit of a strange one, indeed.

I use a debugger to get a grip on other people's code, but if it's my own code, wanting a debugger points to not having thought hard enough.

Printing variables on the other hand is an investment. You evolve to log statements dumping reasonably complete aspects of state, which are disabled in production. If things go bad, you can enable specific loggers even in production on part of your cluster.

This way, troubleshooting in dev won't only fix a bug, it will train your logging to give the relevant info, and only the relevant info.

Now dumping 50K lines of data and hunting specific values in an unstructured way, run after run after run, yeah, that's dumb. You've got a computer, the thing is good at filtering relevant data for you, you might as well use it.

You can pry log debugging from my cold dead hands.

I can justify using a debugger in a backend language like Python.

But it's a pain in the ass for browser javascript that is generated from Typescript/Coffescript or whatever the flavor of the week is.

The value of automating isn't saving a few hours of developer time spread over years. It's improving the developer experience, which is immensely important. Bonus points if the automation removes some human error, such as fat fingering something.

It and the benefits from it cannot be easily measured. But good managers understand its value.

As others note, being forced to context switch is quite damaging. If I have to context switch to press a green button every 30 minutes, sure it only takes 1 second but that's not the point. I've worked on projects that had dozens of these "green buttons".

> Did you know you can get a public URL for a Node application being served from localhost or 127.0.0.1 on your local machine, even if its behind a firewall or corporate network?

Great way to irritate your security team.

> Use Vimium and test your web applications without a mouse

Or just... Test it without a mouse? I don't get why Vimium is recommended here. While its a cool extension, like a fraction of a percent of the users of your web application will be using it.

> For every change it takes 36 seconds before you can get feedback and try out your changes. This might not sound like a lot, but if you are doing this for every change this might happen 50 or more times per day. That’s 1800 seconds, or a half an hour that you could have spent doing other things. > > Over a 5 day working week, this adds up to 25 hours or just over a day of time lost. So if you can solve this problem, it will be as if you worked 6 days a week instead of 5 when all you did is remove a repetitive task.

I think somebody got some wires crossed here. 36 seconds * 50 times per day * 5 days is 9000 seconds or 150 minutes per 5 day work week.

Also, it's not _exactly_ the same, but https://xkcd.com/1205/ comes to mind.

Good work spotting the mistake, this section of the article has been updated with more accurate numbers. Its still a lot of time, something like the equivalent of a working week over a year.

You can also automate multiple tasks, so if you automated 5 tasks that normally take you 36 seconds each you might save yourself 5 weeks in a year.

They lost a decimal and went to 25 hours instead of 2.5 hours. These tips are very generic and something I'd expect any professional to know.

expose reminds me to ngrok. Are they similar services?

Similar, it has pros and cons. The pros is its easier to use and is focused purely on web (http and https) that's also a con because you wouldn't be able to use it for SSH forwarding for example.

Nice, thanks. I'll be bookmarking it to keep it in mind when the task is specifically about HTTP. Easier usage is always a good thing!

Article synopsis: "Use a modern IDE"

Bash scripts? We'll have to agree to disagree. I love that all his other programming recommendations don't work if you're coding bash scripts.

Some good tips here!

Reply on news.ycombinator.com