An open letter
Things I've learned in 20+ years as a software engineer.
When you're deep in the debug zone there's no time for hunt and peckers. Curiously, senior engineers can be a little snooty about touch typing, but anything that gets in the way of expressing yourself can only be a bad thing. You should be able to manage at least 50 wpm.
Eat your own dog food
Use the tools, OS and software you use/write to do a job -- be it your actual day job or your home side projects -- but aim to use your software as an end user. If you target Linux by day then you should be running Ubuntu at night; use Jira at work, then use GitLab Issues to manage your chores. And you need a side-hustle.
Don't just study when somebody thinks you should go on a course. In my whole career I think only one formal training course has had any lasting impact (as it happens it was an introduction to Agile). See if you can convince somebody to pay for a Pluralsight subscription or at least make use of the free trial to pique your interest.
Always. Maybe build a personal website and document what you know. You'll never finish but you have to start.
Make sure you use something that is simple and futureproof (like markdown).
You need a website!
No excuses. You could roll your own but if you're not a web professional it will suck. This is hosted by GitLab Pages.
Develop in the debugger
There are times when you find yourself debugging with
cout but this should be a last resort: it's incredibly slow and doesn't give you anywhere near the insight of a proper debugger. If this is normal then you need to fix your dev environment. See gdbgui and some other Linux debuggers.
Get comfortable describing the complexity of your code.
Even if you don't end up pushing to a repo like GitLab. It's useful to have a lightweight config control tool only a
git init away.
Learn a fundamentally different language
Keep your head in the cloud
Register with a cloud provider, spin up a VM and compile things. Cloud computing isn't going away and just doing this alone exercises a whole heap of excellent transferable skills. In fact you don't even have to provision a VM, just click "Google Cloud Shell" for ad-hoc command line tests.
Give credit where it's due
Make sure to credit your colleagues for any help they give you (probably at standup).
Implement the basics and then optimise
Not my advice but something I'm still reminded of regularly: sometimes a crude implementation is good enough and it will let you get a feel for your new feature's worth before investing additional time.
Programmers waste enormous amounts of time thinking about, or worrying about, the speed of noncritical parts of their programs, and these attempts at efficiency actually have a strong negative impact when debugging and maintenance are considered. We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil. Yet we should not pass up our opportunities in that critical 3%. — Donald Knuth
Use Wireshark regularly. It's quite the rabbit hole if you want to really into it, but just firing it up and opening the IO graph is informative. Create filters for TCP and UDP to get a feel for what traffic you're generating.
If you're really desperate Windows 10 has one built in.
Come up with some equivalent timings in seconds
E.g., opening your laptop takes one second.
|execute typical instruction||1/1,000,000,000 sec = 1 nanosec|
|fetch from L1 cache memory||0.5 nanosec|
|branch misprediction||5 nanosec|
|fetch from L2 cache memory||7 nanosec|
|Mutex lock/unlock||25 nanosec|
|fetch from main memory||100 nanosec|
|send 2K bytes over 1Gbps network||20,000 nanosec|
|read 1MB sequentially from memory||250,000 nanosec|
|fetch from new disk location (seek)||8,000,000 nanosec|
|read 1MB sequentially from disk||20,000,000 nanosec|
|send packet US to Europe and back||150 milliseconds = 150,000,000 nanosec|