Basically almost every Lisp out there has a major downside that turns me off from using it to the extent I consider using Haskell’s s-expression tolerance (like (map odd [1,2,3]) (((.) fst snd) (1,(2,3)))) (which, obviously, would be a very bad practice)

• Common Lisp: Classical OOP everywhere, very bad build automation and packaging (asdf), available implementations have low support availability, bugs are everywhere due to the limited teams who are working on those.
• Scheme: Same as CL, in addition to the scare language support in most editors (not sure even if it has an LSP), also it’s hard to find libraries/packages to build something on.
• EuLisp, ZetaLisp, MacLisp, InterLisp, ISLisp, T, Arc, PicoLisp: I still find it hard to believe those are not in the museum of computing yet.
• Guile, Racket: both look so promising but the lack of interest from community which results in an effect on the number of available packages to build on, makes it much harder to start a new project in any of them without reinventing the wheel.
• Clojure: aside from the JVM dependency and the build system hassle, the language is just too big, still the most plausible option in this whole list though.