Many confused programmers stumbled upon COMP_WORDBREAKS. The documentation states https://www.gnu.org/software/bash/manual/html_node/Bash-Variables.html :
COMP_WORDBREAKS
The set of characters that the Readline library treats as word separators when performing word completion. If COMP_WORDBREAKS is unset, it loses its special properties, even if it is subsequently reset.
On my system:
$ declare -p COMP_WORDBREAKS
declare -- COMP_WORDBREAKS=$' \t\n"\'><=;|&(:'
Why and when if ever is COMP_WORDBREAKS ever useful? Why was it invented, and what is the point of it? Why would ever : or = be breaking words and how did those characters end up there?
I tried browsing the net, but it only resulted in finding out about more confused programmers. The bash changelog just states "COMP_WORDBREAKS now exists" https://github.com/bminor/bash/blob/2bb3cbefdb8fd019765b1a9cc42ecf37ff22fec6/NEWS#L1488 . I was not able to find the reason why COMP_WORDBREAKS exists in the first place.
The question originates because every bash completion function I know of has to workaround COMP_WORDBREAKS by joining the words from CWORD back after COMP_WORDBREAKS has split them. Commonly _comp_get_words or now deprecated __get_cword_at_cursor_by_ref and __reassemble_comp_words_by_ref from bash-completion project are used to workaround COMP_WORDBREAKS. Or, alternatively, COMP_WORDBREAKS is very useful and used in specific contexts when writing bash completion. If so, in which contexts and how is it used?
In contrast, zsh does not have COMP_WORDBREAKS, instead you can execute compset locally from the completion function. What usefulness does COMP_WORDBREAKS brings if it is being set for all completion functions within Bash?