Michal Sojka <sojkam1 at fel.cvut.cz> writes:
> +{
> + local cur prev words cword split
> + _init_completion -s || return
> +
> + $split &&
> + case "${prev}" in
> + --format)
> + COMPREPLY=( $( compgen -W "json sexp text text0" -- "${cur}" ) )
> + return
> + ;;
> + --output)
> + COMPREPLY=( $( compgen -W "sender recipients" -- "${cur}" ) )
> + return
> + ;;
> + --sort)
> + COMPREPLY=( $( compgen -W "newest-first oldest-first" -- "${cur}" )
> )
> + return
> + ;;
> + --exclude)
> + COMPREPLY=( $( compgen -W "true false flag all" -- "${cur}" ) )
> + return
> + ;;
> + esac
> +
> + ! $split &&
> + case "${cur}" in
> + -*)
> + local options="--format= --output= --sort= --exclude="
> + compopt -o nospace
> + COMPREPLY=( $(compgen -W "$options" -- ${cur}) )
> + ;;
> + *)
> + _notmuch_search_terms
> + ;;
> + esac
> +}
> +
I am reminded that we have no tests for shell completion stuff, which
seems pretty fragile.
> +
> + ``--format=``\ (**json**\ \|\ **sexp**\ \|\ **text**\ \|\ **text0**)
> + Presents the results in either JSON, S-Expressions, newline
> + character separated plain-text (default), or null character
> + separated plain-text (compatible with **xargs(1)** -0 option
> + where available).
> +
> + ``--format-version=N``
> + Use the specified structured output format version. This is
> + intended for programs that invoke **notmuch(1)** internally. If
> + omitted, the latest supported version will be used.
> +
I wonder if at some point we should have a notmuch-output-formats.7 page.
> + ``--exclude=(true|false|all|flag)``
> + A message is called "excluded" if it matches at least one tag in
> + search.tag\_exclude that does not appear explicitly in the
> + search terms. This option specifies whether to omit excluded
> + messages in the search process.
Similarly for excludes. I'm ok with the duplication for now, and I can
see an argument for not making the user chase references.
