Many people don't care for it, and I made peace with that many moons ago; personally, I'm quite fond of what I'd call 'applications perl', which at least when I write it tends to be a mixture of OO and function composition type code with as much of the state as possible pushed to the outer edge (plus async for I/O if applicable), etc.
It barely feels like the same language as sysadmin scripts are written in, it just happens to share a compiler and a VM.
... and it doesn't exactly even share syntax since perl supports libraries defining their own keywords, e.g. while I won't be surprised if somebody makes it a feature of the VM eventually, our async/await is currently provided by https://metacpan.org/pod/Future::AsyncAwait and as you can see at https://metacpan.org/pod/Future::AsyncAwait#WITH-OTHER-MODUL... there's quite a few other useful ones out there.
Of course, many people complain vociferously about the idea that you have to use *libraries* to get syntax, but as somebody who still loves scheme I regard the ability to build the language up towards the problem to that extent to be a feature.
My first ever cpan release was making continuations pass back out to perl as a subroutine reference from Guile so I could write linear top level logic in scheme and use continuation capturing to sit that atop an event driven I/O stack - axisofeval.blogspot.com's lispx (https://github.com/lispx/lispx/) provides an fexpr based lisp that does similar for JS.
(trying to think of a good example of perl code of mine that shows an example of this is annoying me because I don't have any public code right now that's new enough for me to not have already decided I don't like it anymore ;)
