closeOnCompletion

Wed Feb 17 19:43:46 UTC 2021

This is indeed what we (postgres driver) are looking for, to know the intent
of closeOnCompletion, this might be a little confusing as the current
wording is a bit unclear about it, but now it's perfectly clear the intent
with the example provided, it would be a nice addition to the spec to
clearly state that use case.

Regards,
Jorge

On Wed, Feb 17, 2021 at 8:26 PM Lance Andersen <lance.andersen at oracle.com>
wrote:

> Hi Filipe,
>
> As Douglas points out at the time of the discussion,  the JDBC EG
> consensus was the Statement would be closed.  If applications did not want
> this behavior they should not call Statement::closeOnCompletion and if they
> were not sure if the method had been invoked, they could always call
> Statement::isCloseOnCompletion to make a decision programatically.
>
> This area as you are aware is messy enough and the intent was to address a
> common use case which lead to common issues for applications and not
> further complicate things.
>
> HTH
>
> Best,
> Lance
>
> On Feb 17, 2021, at 2:14 PM, Filipe Silva <filipe.silva at oracle.com<mailto:
> filipe.silva at oracle.com>> wrote:
>
> Hi Douglas,
>
> Wait, this is not the question here, there is no doubt that closing the
> result sets closes the statement. The question is whether a statement with
> `closeOnCompletion` and an active results set (not closed yet) can be
> re-executed or not. In different words, implicit closing of open result
> sets on next statement execution triggers `closeOnCompletion` and closes
> the statement or should it execute?
>
> Regards,
> Filipe
>
>
> On 17/02/21 19:03, Douglas Surber wrote:
> As I said I wrote the original proposal and what I intended and what the
> EG discussed at the time at least implicitly was that closing the open
> ResultSet would close the Statement and that no subsequent execution was
> possible. No this wasn't spelled out in the discussion. It was clearly the
> intent as the purpose of the method was to handle the case Lance described.
>
> Maybe the language is unclear but I don't think that would justify
> expanding the behavior of the method to cover your use case. The goal is to
> support a specific use case. Any change to the spec should be to more
> clearly state how that use case is supported, not to expand the meaning to
> support additional use cases, especially a use case that has never come up
> in practice.
>
> Douglas
> ________________________________
> From: Filipe Silva <filipe.silva at oracle.com><mailto:
> filipe.silva at oracle.com>
> Sent: Wednesday, February 17, 2021 10:55 AM
> To: Dave Cramer <davecramer at gmail.com><mailto:davecramer at gmail.com>;
> Douglas Surber <douglas.surber at oracle.com><mailto:
> douglas.surber at oracle.com>
> Cc: Lance Andersen <lance.andersen at oracle.com><mailto:
> lance.andersen at oracle.com>; jdbc-spec-discuss at openjdk.java.net<mailto:
> jdbc-spec-discuss at openjdk.java.net> <jdbc-spec-discuss at openjdk.java.net
> ><mailto:jdbc-spec-discuss at openjdk.java.net>
> Subject: Re: closeOnCompletion
>
> Well, looks strange to me. I explain why in the an answer I wrote in
> StackOverflow (
> https://stackoverflow.com/questions/66241480/jdbc-statement-closeoncompletion-should-close-statement-on-next-execution/66248074#66248074
> ).
> Moreover, IMO, allowing re-executions doesn't conflict with the use case
> below and it allows more.
> Regards,
> Filipe
> On 17/02/21 18:38, Dave Cramer wrote:
>
>
> On Wed, 17 Feb 2021 at 13:33, Douglas Surber <douglas.surber at oracle.com
> <mailto:douglas.surber at oracle.com>> wrote:
> I wrote the original proposal. The intent was to handle the case Lance
> described. It was not intended to allow multiple executions of the
> Statement. So while the language may not be as clear as it needs to be, the
> case Filipe described should throw on the second execution of the
> Statement. At least that was my intent.
>
>
> Well there's enough hints in the wording to figure that out, but they are
> just hints.
> We (postgres) are going to implement it in such a way that the statement
> will not be executed twice
>
> Dave
>
> Douglas
> ________________________________
> From: jdbc-spec-discuss <jdbc-spec-discuss-retn at openjdk.java.net<mailto:
> jdbc-spec-discuss-retn at openjdk.java.net>> on behalf of Lance Andersen <
> lance.andersen at oracle.com<mailto:lance.andersen at oracle.com>>
> Sent: Wednesday, February 17, 2021 9:52 AM
> To: Dave Cramer <davecramer at gmail.com<mailto:davecramer at gmail.com>>
> Cc: Filipe Silva <filipe.silva at oracle.com<mailto:filipe.silva at oracle.com>>;
> jdbc-spec-discuss at openjdk.java.net<mailto:
> jdbc-spec-discuss at openjdk.java.net> <jdbc-spec-discuss at openjdk.java.net
> <mailto:jdbc-spec-discuss at openjdk.java.net>>
> Subject: Re: closeOnCompletion
>
> I will have to go dig back through my email archive as this was first
> discussed in 2009.
>
>
> The overall intent was to deal with code similar to:
>
> ----------------------
> ResultSet rs = foo();
>  while(rs.next() {
>      /*do something */
> }
> rs.close();
>
> public ResultSet foo() {
>
>    Statement stmt = con.createStatement();
>    stmt.closeOnCompletion();
>    ResultSet rs = stmt.executeQuery(aQuery);
>    return rs
>
> }
> ------------------
>
>
> We did spend a lot of time on this back in 2009 and took quite a bit of
> time to reach agreement on the current wording.  However, it does  look
> like there is the potential for some additional word smithing.
>
> Best
> Lance
>
>
>
> On Feb 17, 2021, at 12:34 PM, Dave Cramer <davecramer at gmail.com<mailto:
> davecramer at gmail.com><mailto:davecramer at gmail.com<mailto:
> davecramer at gmail.com>>> wrote:
>
> On Wed, 17 Feb 2021 at 12:13, Filipe Silva <filipe.silva at oracle.com
> <mailto:filipe.silva at oracle.com><mailto:filipe.silva at oracle.com<mailto:
> filipe.silva at oracle.com>>> wrote:
>
> On 17/02/21 10:53, Dave Cramer wrote:
> Interesting situation.
>
> If we have an open statement with open resultsets and call
> closeOnCompletion and then execute the statement a second time the
> statement will throw an already closed exception since the spec says that
> executing a statement a second time closes any open resultsets. Closing
> said resultsets ends up closing the statement.
>
> I'm just confirming that this is the intended behaviour. Seems about
> right
> since if you set closeOnCompletion you would not expect to re-use the
> statement.
>
> Dave Cramer
> No, I don't think it is supposed to work like this. "Completion" means
> that the user is done with the ResultSet(s) and explicitely calls
> close() on it/them.
>
> Successive executions do close previously opened ResultSets but closing
> those ResultSets implicitely (by re-executing) do not close the
> Statement. The documentation makes it clear when it says "a call to
> |closeOnCompletion| does effect both the *subsequent execution of
> statements*, (...)". There would be no "subsequent execution of
> statements" if the behavior was the one you described.
>
>
> Well interestingly I'm told that the Oracle driver does close the statement
> on subsequent execution.
>
> Either way I'm not sure I agree with the inference. Now we have to debate
> the word execution. Is calling the method an execution or is completing the
> call an execution ?
>
> Dave Cramer
>
> --
> [Oracle]<http://www.oracle.com/>
> Filipe Silva, Senior Software Developer
> Mobile: +351.91.009.2110
> MySQL Middleware and Clients
> Oracle Portugal, Lagoas Park, Edifício N.º 8, 2740-244 Porto Salvo
>
> --
> [Oracle]<http://www.oracle.com/>
> Filipe Silva, Senior Software Developer
> Mobile: +351.91.009.2110
> MySQL Middleware and Clients
> Oracle Portugal, Lagoas Park, Edifício N.º 8, 2740-244 Porto Salvo
>
>