On Fri, Jun 26, 2009 at 03:47:04PM +0100, Philip Martin wrote:
> Stefan Sperling <stsp_at_elego.de> writes:
>
> > I don't think that opening an access baton like this will work while
> > a commit is in progress. It will work while no commit is in progress,
> > so in your testing the function has suceeded so far.
> >
> > But it will fail during a commit because the root directory of the
> > working copy directory might already be locked for writing.
>
> Read-locks and write-locks can coexist.
Can they? I assumed that readers and readers could co-exist,
but not writers and readers. But I didn't check the code to
see if this is true. Thanks!
Anyway, checking if a path lies within a known WC root, and if it
does not walking downwards from the common root which is not a WC
to find all possible WC roots is still a better plan than trying
to find the root for each target by walking the target path upwards
towards its WC root.
Stefan
Received on 2009-06-26 16:56:48 CEST