Pretty sure Linus never realised his toy operating system would turn into a 30 year project/career.

(I suspect RMS probably suspected the timeframe for GNU, at least for the collection of system utilities rather than the kernel, was going to be a multi decade piece of work.)

Linus was already doing a master’s thesis (like the OP.) He knew Linux wasn’t a weekend project, and he knew the orders of magnitude invoked in creating a commercial OS.

No one can predict the longevity of a project, but that’s not the point.

I would bet good money that if you told Linus in late 1991 that he’d spend his entire career working on his “side project” and that it would be the major OS powering many billion and even trillion dollar new tech companies, and running on several billion battery powered and wireless networked pocket supercomputers in 2021 - and that he’d be some sort of software industry messiah - he’d have laughed at you (and probably viciously evicerated you on a public mailing list for your stupidity, like he is wont to do...)

Hobby projects can be pretty much the same scale as master's thesis, especially for young people.

Implementing RSA is probably the only one i can get behind this advice for. Its easy to screw up, but its also non obvious that you screwed up. If someone actually needs the confidentiality you're trying to provide, the consequences can be far worse than the typical: program does not work.

Other than that, failure is how we learn, and maybe you will succeed.

A charting library is not that difficult.

I honestly can’t tell if you’re joking or not. I mean on one end you can quickly get a toy version up and running, at the other end success looks like: More than 3200 issues open on GitHub. 4,112 commits 22 branches 228 releases 121 contributors

So, this is the perfect example of what I am talking about. Getting something you can generously call a charting library isn’t that hard, getting something worth the time building on it’s own merits is. If you want to have a fun side project go for it. However, if the goal is to actually make charts it’s extremely unlikely that starting from scratch is a good idea.

As a general rule, I don't trust "$thing is not that difficult" statements. Similar to statements of "Just do X", if it's not coming from someone who actually knows what it is that is being done, then I would assume it's more difficult than is being suggested.

I've been writing graphing/charting code off and on for over two decades. My code has been used in high end audio software. Sometimes it's faster and easier to write the simple for loop that draws lines or splines on a canvas than to incorporate a "full-featured" charting library that has to support every kind of chart, plot, graph, network, 2D, 3D, etc. imaginable. A charting library that takes an array and makes lines is not that hard.

And yet I still had a dev get pissed off by my offer to help with a charting problem. "Yeah, but was it D3?"

I'm generally an advocate for buy-before-build in software, but I don't trust the kind of dev who won't even acknowledge that the easy part of a problem is easy (or that true experience isn't framework specific).

Indeed.

A couple of months ago, a junior developer mentioned he'd need a library for a bar chart, and asked for directions how to use it in an Xcode project. It turned out to be easier to draw the chart with a few rectangles, and the developer had one of those great a-ha moments.