"J.P." writes: > Something else not included that perhaps should be is a clearer > indication that items added interactively can match any substring in a > candidate without regard for boundaries. It may also be worth > contrasting this with what's ultimately highlighted. For example, if you > "M-x erc-add-pal RET bo RET", then "" (minus the brackets) will be > highlighted. Compare this to keywords, where only the matched portion > appears in a designated face. This is nonsense (apologies). ERC already includes this information in the doc strings for `erc-current-nick-highlight-type' and friends. > Documentation aside, it might also be nice to retain a means of > interactively adding a regexp verbatim (IOW, the old behavior minus the > unhelpful completion list). Perhaps this can be done via universal > argument. I've added an option offering to quote new items, do nothing, or ask when applicable. By default, it's set to the latter to accommodate infrequent users (those likeliest to have been bitten by the old behavior). This may be received negatively by some, but there's already precedent in `erc-cmd-IGNORE', which has no accompanying opt-out. Also, when the choice is quote or verbatim, the opposite boolean is made available via universal argument. For example, if you're a fan of the status quo because you like manually composing regexps on the fly (and don't mind slightly misleading completion offerings), you may take comfort in knowing that ERC will happily quote them for you when supplied with a prefix argument, should the unlikely need ever arise. > BTW, the individual who originally brought this to our attention on > Libera also mused about a command for jumping between keywords, so I've > included something sketchy to that effect. However, it suffers from an > intermittent issue that I can't yet reproduce reliably: basically, when > first invoked, it sometimes makes Emacs unresponsive for a few seconds. > If I can't manage to tame it, I'll just drop it from this patch set. Despite my weird issue having magically subsided after syncing and rebuilding, I have, for now, decided to drop the proposed command, mainly because I've been reminded of a discussion [1] from a few months back regarding a proposal for a mini framework for jumping between elements of any type. I'd rather save this functionality for something like that, provided such a thing ever materializes. Thanks. [1] https://lists.gnu.org/archive/html/emacs-erc/2022-02/msg00010.html