Re: [eclipselink-dev] JPA 2.0: Ordered Lists open issues

[Gordon Yorke <gordon.yorke@xxxxxxxxxx>]

I agree that we shouldn't be throwing exceptions on read, but silently 
allowing this error case to pass is also bad.  Using nulls for missing 
indexes on reads allows us to show the user where the problem is and 
allows us to raise an error on commit if the null is still present in 
the collection. 
--Gordon

Douglas Clarke wrote:
> Andrei,
>
> I do not believe throwing an exception when non-contiguous index values are found is an option. That would mean a developer using EclipseLink would have to wrap every collection mapping access in a try-catch block. Ultimately we need to support reading these in and writing something back out if the list is changed.
>
> I also do not believe missing values should be interpreted as null entries in the collection.
>
> As far as fixing the index values go we could reset them all if any changes are made or we deal with the more complex solution of figuring out values for new and changed values without a broader update to all rows.
>
> Doug
>
> -----Original Message-----
> From: Andrei Ilitchev 
> Sent: Monday, February 09, 2009 11:32 AM
> To: Dev mailing list for Eclipse Persistence Services
> Subject: [eclipselink-dev] JPA 2.0: Ordered Lists open issues
>
>
> I've added Requirements paragraph to Ordered Lists Wiki page containing Open 
> Issues, please take a look
> http://wiki.eclipse.org/EclipseLink/Development/JPA_2.0/ordered_lists
>
> Thanks,
>
> Andrei 
>
> _______________________________________________
> eclipselink-dev mailing list
> eclipselink-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/eclipselink-dev
> _______________________________________________
> eclipselink-dev mailing list
> eclipselink-dev@xxxxxxxxxxx
> https://dev.eclipse.org/mailman/listinfo/eclipselink-dev
>