[Dbix-class] Path roundup, mark II

luke saunders luke.saunders at gmail.com
Fri Apr 25 18:38:10 BST 2008


On Fri, Apr 25, 2008 at 5:13 PM, Peter Rabbitson <rabbit+list at rabbit.us> wrote:
> Matt S Trout wrote:
>
> > On Wed, Apr 23, 2008 at 10:20:16PM +0200, Peter Rabbitson wrote:
> >
> > > Matt S Trout wrote:
> > >
> > > > Anything left or do we start cutting 08100 dev releases?
> > > >
> > > > (the "single accessor doesn't return undef" test patch doesn't count
> > > > unless somebody produces one that doesn't break the rest of the test
> suite,
> > > > it's a silly bug but I'm not delaying for it)
> > > >
> > > >
> > > There was some crazyness that leaked from SQLT, don't know if it can be
> considred serious enough (makes deploy of moderately complex schemas
> unusable on mysql, possibly others):
> http://lists.scsys.co.uk/pipermail/dbix-class/2008-April/006056.html
> > >
> >
> > Is that our problem or SQLT's?
> >
>
>  I'd think DBIC since the affected file is part of the DBIC distribution.
>
>
>
> > If ours, is anybody interested in writing a failing test and/or a patch
> for
> > this?
> >
>
>  There is nothing to write really, just a reversal of the patch introducing
> the regression (mentioned in the email thread). Nobody ever claimed the
> commit, so I am not sure what the procedure is. If you release the way it is
> now deploy() will be seriously borked on mysql at the very least. I can
> write tests for this if you think it will be beneficial (so it doesn't leak
> in again).
>
>  OTOH I have to admit that I am not entirely sure what was the purpose of
> the initial changes. Seems very counterintuitive.
>

It was me committed it originally. There was a reason for it which
involved diffs being screwy if the database auto-generated indexes
weren't explicitly specified in the DDLs, but I agree that this fix
isn't sane - if anything the fix should be in the SQLT producers.

I have removed the extra indexes for now and will address the problem
properly should it come up again.

http://dev.catalystframework.org/svnweb/bast/revision/?rev=4297

And some parser tests would be great if you could.



More information about the DBIx-Class mailing list