Re: argmatch: accept perfect matches in documented arglists

[Akim Demaille]

> Le 27 avr. 2019 à 14:33, Bruno Haible <address@hidden> a écrit :
> 
> Hi Akim,

Hi Bruno,

> Sorry but I don't understand this:
> 
>> $ bison --trace=help
>> bison: invalid argument ‘help’ for ‘--trace’
>> Valid arguments are:
>>  - ‘none        - no traces’
>>  - ‘scan        - grammar scanner traces’
>>  - ‘parse       - grammar parser traces’
>>  - ‘parse-stats - grammar parser stats’
>>  - ‘automaton   - construction of the automaton’
>>  - ‘bitsets     - use of bitsets’
>>  - ‘closure     - input/output of closure’
>>  - ‘grammar     - reading, reducing the grammar’
>>  - ‘resource    - memory consumption (where available)’
>>  - ‘sets        - grammar sets: firsts, nullable etc.’
>>  - ‘muscles     - m4 definitions passed to the skeleton’
>>  - ‘tools       - m4 invocation’
>>  - ‘m4          - m4 traces’
>>  - ‘skeleton    - skeleton postprocessing’
>>  - ‘time        - time consumption’
>>  - ‘ielr        - IELR conversion’
>>  - ‘all         - all of the above’
> 
> Why not:
> 
> $ bison --trace=help
> bison: invalid argument ‘help’ for ‘--trace’
> Valid arguments are:
>  - ‘none’        - no traces
>  - ‘scan’        - grammar scanner traces
>  - ‘parse’       - grammar parser traces
>  - ‘parse-stats’ - grammar parser stats
>  - ‘automaton’   - construction of the automaton
>  - ‘bitsets’     - use of bitsets
>  - ‘closure’     - input/output of closure
>  - ‘grammar’     - reading, reducing the grammar
>  - ‘resource’    - memory consumption (where available)
>  - ‘sets’        - grammar sets: firsts, nullable etc.
>  - ‘muscles’     - m4 definitions passed to the skeleton
>  - ‘tools’       - m4 invocation
>  - ‘m4’          - m4 traces
>  - ‘skeleton’    - skeleton postprocessing
>  - ‘time’        - time consumption
>  - ‘ielr’        - IELR conversion
>  - ‘all’         - all of the above

Because that's not what argmatch provides.  You seem to be referring to
a fully blown feature, and support for internationalization I guess.
I'm just exploiting the unique prefix feature to shoehorn the
documentation in there.

>> +   Accept arguments with a documentation string, such as:
>> +     "foo - a first feature" -> 0
>> +     "bar - a second feature" -> 1
> 
> It doesn't make sense to me to expect that the user will type
>  bison --trace='none        - no traces'
> when they can type
>  bison --trace=none

Yes, of course not.  No one would ever do that.  Then again, this
option is not even documented.