Fwd: Completion confusion around the _arguments -A flag

Fwd: Completion confusion around the _arguments -A flag

[Mal Graty]

[-- Attachment #1: Type: text/plain, Size: 1284 bytes --]

I have a binary (call it foo) with two, mutually exclusive, optional flags
(that must come first) and a required arg. The problem I'm having is that
when omitting both flags and using just the arg the completion tries to add
one of the flags after it. Reading the docs lead me to the -A flag, however
it doesn't seem to be preventing the flag completion despite being after
the first arg. I'm hoping I've just overlooked something but for the life
of me I can't see it.

Command structure:

foo [--name|--software] <node>

Current completion code:

#compdef foo

_nodes() {
  compadd $(get_node_list)
}

local curcontext="$curcontext" state state_descr line
typeset -A opt_args

local rc=1

_arguments -A '-*' \
  '(--name --software)'{--name,--software} \
  '1: :_nodes' \
  && rc=0

return rc

Examples:

foo <TAB>              #=> GOOD: nodelist
foo -<TAB>             #=> GOOD: --name --software
foo --name <TAB>       #=> GOOD: nodelist
foo --name -<TAB>      #=> GOOD: nothing
foo --software <TAB>   #=> GOOD: nodelist
foo --software -<TAB>  #=> GOOD: nothing
foo node <TAB>         #=> BAD: --name --software, expect nothing
foo node -<TAB>        #=> BAD: --name --software, expect nothing

Any assistance to solve this issue would be massively appreciated!

Many thanks,
Mal

Re: Completion confusion around the _arguments -A flag

[dana]

On 22 Jul 2018, at 10:16, Mal Graty <mal@idio.ai> wrote:
>I have a binary (call it foo) with two, mutually exclusive, optional flags
>(that must come first) and a required arg. The problem I'm having is that
>when omitting both flags and using just the arg the completion tries to add
>one of the flags after it. Reading the docs lead me to the -A flag, however
>it doesn't seem to be preventing the flag completion despite being after
>the first arg. I'm hoping I've just overlooked something but for the life
>of me I can't see it.

For whatever it's worth, i don't see anything obviously wrong with your
function, and it works fine when i test it on 5.5. Maybe you have an option
enabled that changes the behaviour, idk

dana

Re: Completion confusion around the _arguments -A flag

[Mal Graty]

[-- Attachment #1: Type: text/plain, Size: 1056 bytes --]

Thanks Dana

For reference, in case it rings a bell for anyone, I'm seeing this on a
vanilla Ubuntu 18.04 install. I'll have a poke around for anything that
might be altering default behaviour though, thanks for confirming I'm
not mad :)

On 25 July 2018 at 17:36, dana <dana@dana.is> wrote:

> On 22 Jul 2018, at 10:16, Mal Graty <mal@idio.ai> wrote:
> >I have a binary (call it foo) with two, mutually exclusive, optional flags
> >(that must come first) and a required arg. The problem I'm having is that
> >when omitting both flags and using just the arg the completion tries to
> add
> >one of the flags after it. Reading the docs lead me to the -A flag,
> however
> >it doesn't seem to be preventing the flag completion despite being after
> >the first arg. I'm hoping I've just overlooked something but for the life
> >of me I can't see it.
>
> For whatever it's worth, i don't see anything obviously wrong with your
> function, and it works fine when i test it on 5.5. Maybe you have an option
> enabled that changes the behaviour, idk
>
> dana
>
>