Re: [bug] attributes stuff in options_menu.c broken

Sorry for the confusion, I have to correct me again.

On 01.11.2008, at 09:47, Dennis Preiser wrote:
>> currently I'm incorporating tinrc-variables into the attributes  
>> system. During testing I ran into trouble with the  
>> options_menu.c:change_config_file()-code.
>>
>> The following lines:
>>
>> group->attribute->ATTRIBUTE = tinrc.ATTRIBUTE;
>> glob_attributes.ATTRIBUTE = tinrc.ATTRIBUTE;
>>
>> are useless in my mind.
>>
>> If you change an option at group-level, the first line overwrites  
>> a attribute set from the attributes-file (if present).

This:

>> If you change an option at group-level, the second line causes tin  
>> to write this changed value to all scopes which matches this group  
>> (if this option is not set in the attributes-file for those scopes).

is nonsense and should be:

If you change an option at group-level and this option is not set in  
the attributes-file for this group, the second line causes tin to  
write the *former* value of this option to all scopes which matches  
this group and which have not set this option already.

>> I'm in favor of wiping out all attributes-specific stuff from  
>> options_menu.c:change_config_file() and call read_attributes_files 
>> () in options_menu.c this way:

This:

>> case GLOBAL_QUIT:
>> 	write_config_file(local_config_file);
>> 	read_attributes_files();
>> 	break;

is nonsense too and should be:

case GLOBAL_QUIT:
	write_config_file(local_config_file);
	read_attributes_files();
	/* FALLTHROUGH */
case CONFIG_NO_SAVE:
	clear_note_area();
	return;

in options_menu.c:change_config_file().

I have used the latest snapshot. The option has to be available as an  
attribute to reproduce the described behavior (for instance  
trim_article_body).

Dennis