Re: [Qemu-devel] [PATCH 3/7] monitor: Add migrate_set_capability completion.

[Markus Armbruster]

"Dr. David Alan Gilbert" <address@hidden> writes:

> * Hani Benhabiles (address@hidden) wrote:
>> Signed-off-by: Hani Benhabiles <address@hidden>
>> ---
>>  hmp-commands.hx |  1 +
>>  hmp.h           |  2 ++
>>  monitor.c       | 21 +++++++++++++++++++++
>>  3 files changed, 24 insertions(+)
>> 
>> diff --git a/hmp-commands.hx b/hmp-commands.hx
>> index 45e1763..919af6e 100644
>> --- a/hmp-commands.hx
>> +++ b/hmp-commands.hx
>> @@ -975,6 +975,7 @@ ETEXI
>>          .params     = "capability state",
>>          .help       = "Enable/Disable the usage of a capability for 
>> migration",
>>          .mhandler.cmd = hmp_migrate_set_capability,
>> +        .command_completion = migrate_set_capability_completion,
>>      },
>>  
>>  STEXI
>> diff --git a/hmp.h b/hmp.h
>> index a70804d..0c814d0 100644
>> --- a/hmp.h
>> +++ b/hmp.h
>> @@ -107,5 +107,7 @@ void ringbuf_write_completion(ReadLineState *rs, int 
>> nb_args, const char *str);
>>  void ringbuf_read_completion(ReadLineState *rs, int nb_args, const char 
>> *str);
>>  void watchdog_action_completion(ReadLineState *rs, int nb_args,
>>                                  const char *str);
>> +void migrate_set_capability_completion(ReadLineState *rs, int nb_args,
>> +                                       const char *str);
>
> Thank you for doing this; I spend way too much time typing these commands.
>
>>  #endif
>> diff --git a/monitor.c b/monitor.c
>> index fb300c2..6a3a5c9 100644
>> --- a/monitor.c
>> +++ b/monitor.c
>> @@ -4572,6 +4572,27 @@ void watchdog_action_completion(ReadLineState *rs, 
>> int nb_args, const char *str)
>>      add_completion_option(rs, str, "none");
>>  }
>>  
>> +void migrate_set_capability_completion(ReadLineState *rs, int nb_args,
>> +                                       const char *str)
>> +{
>> +    size_t len;
>> +
>> +    len = strlen(str);
>> +    readline_set_completion_index(rs, len);
>> +    if (nb_args == 2) {
>> +        int i;
>> +        for (i = 0; i < MIGRATION_CAPABILITY_MAX; i++) {
>> +            const char *name = MigrationCapability_lookup[i];
>> +            if (!strncmp(str, name, len)) {
>> +                readline_add_completion(rs, name);
>> +            }
>> +        }
>> +    } else if (nb_args == 3) {
>> +        add_completion_option(rs, str, "on");
>> +        add_completion_option(rs, str, "off");
>> +    }
>
> It's a shame you have to do all of these manually; if we could tell something
> that we had an enum of 'MigrationCapability' then it could remove the command
> specific glue.

HMP command arguments are specified by an args_type string, which can't
express "enumeration", so we use strings instead.

We could hack in another arguments type 'E', abusing the parameter name
like type 'O'.  But I think the args_type string is long past its "best
before" date.

We could replace it by a command schema, like QMP's.  Differences to QMP
include: arguments are positional, fewer types are accepted, and there's
convenience syntax such as size suffixes.

Not exactly an easy task, I'm afraid.

[...]