There are Usenix papers on bash and ksh -- I archived the ksh one here: http://www.oilshell.org/archive/ksh-usenix.pdf http://www.oilshell.org/archive/ksh-usen

There are Usenix papers on bash and ksh -- I archived the ksh one here: http://www.oilshell.org/archive/ksh-usenix.pdf http://www.oilshell.org/archive/ksh-usenix.pdf The surprising thing to me is that ksh was Lua-like in terms of being embeddable and Perl-/Python-like in terms of having GUI bindings(!) and being a rich language. Though I don't think it has garbage collection and thus was fundamentally more limited in its data structures than Perl/Python/JS/Ruby/PHP, etc. My viewpoint is that shell should be more like the dynamic languages that won -- which are basically those 5. I elaborated on that here: https://old.reddit.com/r/ProgrammingLanguages/comments/n1y2pd/lowlevel_and_sandboxed_scripting_all_in_a_single/gwkcl4l/ https://old.reddit.com/r/ProgrammingLanguages/comments/n1y2p... Probably one of the longest documents is a 182 page Master's thesis from the author of the Oh shell, linked here: https://github.com/oilshell/oil/wiki/Alternative-Shells https://github.com/oilshell/oil/wiki/Alternative-Shells There are many sections, but one of them goes into history, and other sections explore alternatives. ----- Ken Thompson's original paper on shell just popped up: https://news.ycombinator.com/item?id=25235450 https://news.ycombinator.com/item?id=25235450 (it was scanned and revived) There are also interviews with Stephen Bourne around. I would say that those 4 shells form the "dominant" strain: Thompson shell -> Bourne Shell -> ksh -> bash, although there are many offshoots. ---- If you want some fun reading that's not about the shell specifically but about Unix and language-oriented interfaces, check out "In the Beginning Was the Command Line" by Neal Stephenson (search HN archives, it has appeared many many times). Also I recommend "The Art of Unix Programming" by Raymond for how to design programs that work well with shell. It has history and Ken Thompson gives some commentary (in the print version, at least). ---- I have a lot of thoughts on where shell is going in the future and I try to put most of them on https://www.oilshell.org/ https://www.oilshell.org/ ! Although it's easy to design something in your head but often hard to implement a solid version of it. For example I still think shell, awk, and make should be combined but that was over 4 years ago :) http://www.oilshell.org/blog/2016/11/13.html http://www.oilshell.org/blog/2016/11/13.html (I have made some progress on those fronts by deferring some tasks to Ninja, which is very Unix-y and nice. And some of the foundations for awk-like structured data are now in Oil)