And those Python and Java ecosystems are the result of years of effort from their respective communities. When I first started learning Python, the question people would raise was “Why not just use Perl? cpan has just about anything you could want!”
A language needs to be good enough for developers to enjoy using it. Java was a breath of fresh air compared to 90s C/C++ code, so people were very motivated to build with it. Likewise, Python had a reputation for “Looking like pseudo code” which made it attractive to devs who needed a language that would let them build out ideas quickly.
> A language needs to be good enough for developers to enjoy using it.
That's too far. A language simply needs to not be actively bad.
Perl, unfortunately, was actively bad because everybody had their own subset of Perl that they knew and none of them overlapped.
When I was interveiwing for VLSI positions back in the mid 1990's, everybody would ask if you knew Perl. I would start whiteboarding and invariably would write something in Perl that the interviewer wouldn't know--which would effectively derail the interview. This is actively bad.
Python, by constrast, never had that problem. I used to carry to interviews printouts of a reasonably simple program in both Perl and Python (verilog VCD parser) so I could discuss them with the interviewer. The interviewer would always like the Python program better even if they didn't know Python.
(although, to be fair, it seems that Python has now grown that subsetting problem recently--how times change ...)
A language needs to be good enough for developers to enjoy using it. Java was a breath of fresh air compared to 90s C/C++ code, so people were very motivated to build with it. Likewise, Python had a reputation for “Looking like pseudo code” which made it attractive to devs who needed a language that would let them build out ideas quickly.