Donnie Berkholz wrote:
> On 10:11 Mon 01 Oct     , Alin Năstac wrote:
>   
>> In this case, $(use mysql || use postgres && use_enable virtual-users)
>> will result in use_enable virtual-users being executed if and only if
>> USE="-mysql postgres". See this pseudocode:
>>     
>
> No .... if use mysql succeeds, || use postgres is skipped entirely and 
> we go straight to the && use_enable. It's considered directly in order 
> with no precedence.
>
>   
You're right again. :-[
>>     if use mysql do nothing
>>     else if use postgres
>>           use_enable virtual
>>
>> Wouldn't be best to die in pkg_setup if USE="virtual-users -mysql
>> -postgres" ?
>>     
>
> It's better to do something smart than to force interactivity. In this 
> case, you as the maintainer would decide whether this USE combo meant 
> the user wanted virtual-users (if so, forcibly enable one of mysql or 
> postgres), or whether the user definitely did not want to pull in a 
> database (if so, force off virtual-users).
>   
I cannot force mysql or postgres dependency just because user enabled
virtual-users flag.
IMO this case would definitely qualify for a pkg_setup die because
virtual-users is a local useflag. By selecting virtual-users without
enabling mysql or postgres storage backend, the user proved to be
clueless and ebuild should provide the much needed clue.

Attachment: signature.asc
Description: OpenPGP digital signature

Reply via email to