From time to time I go back to some coding challenges but I often struggle with “easy” ones.
At the same time I’m working on a personal toy language compiler (coded in Crystal of course) which is far far more complex than any of the “easy”, even “intermediate” challenges.
I won’t say I’m doing anything special, that’s just a pet language but until now it’s been more or less “ok” (I won’t say easy) to write the tokenizer, the ast builder, the rules checker and the IR generator.
And even if I’m now struggling to death coding the backends for some vintage CPUs how can I be stuck on a “please transpose this matrix” ?
They just require different skills. Some people are more competent in transposing (btw inplace transposition of rectangular matrix is not an easy task, you can check wikipedia about it), others prefer compilers, others prefer raytracers on vulkan, others can do all three areas easily but work as devops etc.
I feel no problems in medium problems on leetcode, but won’t even try to asm on vintage cpu or something like this.
In defense of coding challenges - they are more about knowledge of wide range of algorithms and approaches while personal projects are more about deeply knowing one area under one angle of view. So some practice in algorithms could be useful in your personal project in an unpredictable way.
A lot of people argue that the better way to improve your skills is by coding real things which I always do (coding, about improving I don’t know).
I left the “matrix” challenge in favor of a BST one, managed to pass it but at the end
I don’t need BST for now
I’ll forget what I did in a matter of days
I won’t say that what I feel is right, nor it can apply to everybody (maybe people that want to be hired use coding challenges to prepare themselves for interviews), but at the end I would better spend time coding a vintage PacMac clone and do some researches about the needed algorithms
Sorry for the noise, but hey we’re in “off topic” ;-)