Re: [PATCH] r25137 causes compile failures if sasl is available

Kamesh Jayachandran wrote:
> [[[
>
> r25137 breaks sasl builds because of 'svn_atomic__init_once' signature change.
>
> * subversion/include/private/ra_svn_sasl.h
> (svn_ra_svn__sasl_common_init): Accepts 'pool'.
>
> * subversion/libsvn_ra_svn/ra_svn.h
> (svn_ra_svn__sasl_init): Accepts 'pool'.
>
> * subversion/svnserve/server.h
> (cyrus_init): Accepts 'pool'.
>
> * subversion/libsvn_ra_svn/cyrus_auth.c
> (svn_ra_svn__sasl_common_init): Create sasl_pool from the passed in pool.
> (sasl_init_cb): Accepts 'pool'. Passes the pool to
> 'svn_ra_svn__sasl_common_init'
> (svn_ra_svn__sasl_init): Accepts 'pool'. Passes the pool to
> 'svn_atomic__init_once'
>
> * subversion/libsvn_ra_svn/client.c
> (svn_ra_svn__init): Passes 'pool' to 'svn_ra_svn__sasl_init'.
>

Hi Kamesh,

I think it's best not to make sasl_pool dependent on the lifetime of the
pool that was passed to svn_ra_svn__init. So, instead of revving
svn_ra_svn__sasl_init to take a pool parameter, just pass NULL to
svn__atomic_init_once.

The svnserve changes should be fine, because in that case the pool is
long-lived.

-- 
Vlad
---------------------------------------------------------------------
To unsubscribe, e-mail: dev-unsubscribe@subversion.tigris.org
For additional commands, e-mail: dev-help@subversion.tigris.org