On Fri, Oct 14, 2011 at 10:11:53AM +0200, Stefan Fuhrmann wrote:
> The current code operates along two dimensions:
> * provide the same interface whether or not
> APR provides synchronization structures
> * allow for the application to decide _dynamically_
> whether it is actually necessary to create and
> use a mutex (see svn_cache__t API for instance).
>
> If APR_HAS_THREADS is 0 and the application
> insists on enabling a mutex, svn_mutex__init() will
> return an error.
>
> So, maybe we should redefine the parameter to
> mean, "_if_ I am a multi-threaded application, do I
> need a mutex here". Then it could always be set
> to TRUE or FALSE and the APR_HAS_THREADS
> setting would only be evaluated within *__init().
>
> I prepared a patch that does just that and fixes up
> all callers as well.
Sounds good to me. Thanks!
Received on 2011-10-14 12:33:33 CEST