On Thu, 04 Jan 2007, Malcolm Rowe wrote:
> On Thu, Jan 04, 2007 at 11:20:24AM -0800, Daniel Rall wrote:
> > I'm not sure about this. I like the fact that 'svnadmin create'
> > attempts to create the sqlite merge info index, and think that it
> > should continue to do so, on the grounds that it provides a "fail
> > fast" behavior.
>
> There's arguments for both ways, I think. Creating the index on-demand
> means that we get the same code path and behaviour for repositories that
> are upgraded and those that are created with 1.5, for example.
>
> But why not remove the explicit creation logic and just exercise the
> on-demand code at creation time (by running a dummy query). That way,
> you get the benefit of failure at repository creation time together with
> a single code path to test.
>
> Alternatively, keep the explicit creation - it's really not that much
> code.
So long as 'svnadmin create' continues to create the merge info index,
either implementation sounds okay to me.
- application/pgp-signature attachment: stored
Received on Sat Jan 6 02:08:14 2007