[jsr-221-eg] SQL 2016 MATCH RECOGNIZE JDBC parameter Marker / Escape Characters

[Douglas Surber]

Oracle raised this issue before we added ‘{\ .. \}’ and tried our best to get the EG to agree on something, anything. We proposed ‘{\ . . . \}’ but were abundantly clear that we would accept anything that the EG would agree to. Unfortunately the EG did not come to any agreement. 

Oracle Database added MATCH RECOGNIZE to Oracle SQL and Oracle Database JDBC had to do something. When the EG did not agree to anything we went with what we thought was the best solution.

We did our best to include the wider community when this issue first arose in 2013. The community did not act so Oracle Database JDBC did what we had to to support Oracle Database. I’m sorry that things turned out that way. But now that Oracle Database JDBC has been supporting this syntax for five years, I am going to push to make the established syntax the standard. I absolutely agree that this is not the best way to do things but we tried our best to get some kind of agreement first.

Douglas



> On Oct 9, 2018, at 9:02 AM, Mark Rotteveel <mark at lawinegevaar.nl> wrote:
> 
> On 9-10-2018 17:30, Douglas Surber wrote:
>> Oracle Database JDBC has supported ‘{\ … \}’ as an escape sequence since 2013. 
> 
> Don't take this the wrong way, but I can't help but notice that it looks like Oracle regularly makes its own interpretation of the JDBC specification(*), or does things that are later added to JDBC under the guise of "we already do this in Oracle", instead of discussing it upfront.
> 
> Oracle may be the steward of JDBC, but it is not their personal playground.
> 
> This is especially troubling because a similar problem was discussed back in 2013 for PostgreSQL, and the response back then was along the lines of "there is nothing in JDBC or SQL, so you're on your own", and now it turns out that Oracle invented its own syntax around that same time and now wants to standardize it.
> 
> This could and should have been raised back then, and it should have been discussed in the expert group or on spec-discuss before it was implemented, even if only to avoid different drivers having to invent their own syntax.
> 
> You're now essentially forcing us to accept your syntax, because that is what you already have (or we end up with https://xkcd.com/927/).
> 
> *: eg see https://community.oracle.com/message/13952214
> 
> -- 
> Mark Rotteveel