On Fri, Mar 29, 2013 at 8:04 PM, Daniel Shahaf <danielsh_at_elego.de> wrote:
> Stefan Fuhrmann wrote on Fri, Mar 29, 2013 at 11:45:23 +0100:
> > On Fri, Mar 29, 2013 at 7:19 AM, Daniel Shahaf <danielsh_at_elego.de>
> wrote:
> >
> > > I have a function that looks like this:
> > >
> > > static svn_error_t *f(svn_fs_t *fs)
> > > {
> > > svn_fs_t *ft = svn_fs_open(fs);
> > > svn_fs_verify_root(ft, r42);
> > > }
> > >
> > > I'd like the 'verify' call to bypass any information that is already
> > > cached: I don't mind if it the 'verify' process caches information for
> > > itself and then reads it back, but I don't want it to read information
> > > that is _right now_ in the cache (due to FS, or some other handle to
> the
> > > same filesystem in another thread) in lieu of parsing it, for itself,
> > > from the revision file.
> > >
> > > What would be the best way to achieve this? Do I set every cache in
> > > ft->ffd->* to NULL or to a pointer to a newly-created, empty cache
> object?
> > > Something else?
> > >
> >
> > Originally, I had planned to introduce cache namespaces in 1.9
> > because all processes would share the same cache memory
> > (so, svnadmin would then need use a separate key space).
> >
> > But since you obviously needed right now, I added the feature
> > now in 1462436 on /trunk. Since I changed your code from
> > r1462413, please have a look at my commit to ensure that it
> > still does what you intended.
> >
>
> Looks good to me, many thanks!
>
> One comment: the new docstring should forbid ":" from being used in the
> namespace name.
>
r1462907 addresses this issue in a more generalized way.
-- Stefan^2.
--
*Join one of our free daily demo sessions on* *Scaling Subversion for the
Enterprise <http://www.wandisco.com/training/webinars>*
*
*
Received on 2013-03-31 12:04:20 CEST