Rename COLL_* aggregate functions into STRICT_*

[Till Westmann <ti...@apache.org>]

Hi,

There are 2 kinds of aggregation functions in SQL++ right now:
1) those with the prefix ARRAY_ (which share SQL's behavior or ignoring 
NULL) and
2) those with the prefix COLL_ (which return NULL, if one of the values 
is NULL).

To reduce the confusion about which functions implement a strict NULL 
handling I’ve renamed the COLL_ functions to use the prefix STRICT_ 
instead. To ease migration both prefixes are allowed right now.
But I’d like to get rid of the COLL_ prefix after the 0.9.4 release.

Would that work (could we migrate current uses of the COLL_ prefix 
relatively quickly)?
Or do we know of uses of these functions that cannot easily be migrated?

Cheers,
Till

[1] https://issues.apache.org/jira/browse/ASTERIXDB-2399
[2] https://asterix-gerrit.ics.uci.edu/#/c/2691

Re: Rename COLL_* aggregate functions into STRICT_*

[Mike Carey <dt...@gmail.com>]

+1 for the quick approach.  I doubt that we have a big legacy in this 
part of the SQL++ space...  :-)


On 6/8/18 9:15 PM, Till Westmann wrote:
> Hi,
>
> There are 2 kinds of aggregation functions in SQL++ right now:
> 1) those with the prefix ARRAY_ (which share SQL's behavior or 
> ignoring NULL) and
> 2) those with the prefix COLL_ (which return NULL, if one of the 
> values is NULL).
>
> To reduce the confusion about which functions implement a strict NULL 
> handling I’ve renamed the COLL_ functions to use the prefix STRICT_ 
> instead. To ease migration both prefixes are allowed right now.
> But I’d like to get rid of the COLL_ prefix after the 0.9.4 release.
>
> Would that work (could we migrate current uses of the COLL_ prefix 
> relatively quickly)?
> Or do we know of uses of these functions that cannot easily be migrated?
>
> Cheers,
> Till
>
> [1] https://issues.apache.org/jira/browse/ASTERIXDB-2399
> [2] https://asterix-gerrit.ics.uci.edu/#/c/2691