[PATCH 1/5] log: Refactor evaluation of log level

[Torsten Hilbrich]

Am 29.05.2017 um 13:27 schrieb Aleksander Morgado:
> Hey Torsten,
> 
> On 22/05/17 07:49, Torsten Hilbrich wrote:
[...]
>> diff --git a/src/mm-log.c b/src/mm-log.c
>> index bac0fce..e0255c3 100644
>> --- a/src/mm-log.c
>> +++ b/src/mm-log.c
>> @@ -65,6 +65,22 @@ static const LogDesc level_descs[] = {
>>  static GString *msgbuf = NULL;
>>  static volatile gsize msgbuf_once = 0;
>>  
>> +static int
>> +mm_to_syslog_priority(int level)
> 
> A whitespace is missing before the "(".
> 
> Also, I'm thinking we could actually setup a typedef enum for the MM logging level, something like 
> 
> typedef enum {
>     MM_LOG_LEVEL_ERR   = 0x00000001,
>     MM_LOG_LEVEL_WARN  = 0x00000002,
>     MM_LOG_LEVEL_INFO  = 0x00000004,
>     MM_LOG_LEVEL_DEBUG = 0x00000008
> } MMLogLevel;
> 
> What do you think?

This is a nice idea. I will check what changes are needed elsewhere in
the code and prepare an extra patch for this.

> 
> I'm suggesting this because this would let us pass "MMLogLevel level" in this new method, which the compiler will treat in a more strict way when checking for missing items in the switch.
> 
>> +{
>> +    switch (level) {
>> +    case LOGL_DEBUG:
>> +        return LOG_DEBUG;
>> +    case LOGL_WARN:
>> +        return LOG_WARNING;
>> +    case LOGL_INFO:
>> +        return LOG_INFO;
>> +    case LOGL_ERR:
>> +        return LOG_ERR;
>> +    }
>> +    return LOG_INFO;
> 
> And if we add the new typedef, this fallback return LOG_INFO would also go away, and we can set here g_assert_not_reached().

Will change that too.

	Torsten