Re: FDW-based dblink (WIP) - Mailing list pgsql-hackers

From Itagaki Takahiro
Subject Re: FDW-based dblink (WIP)
Date
Msg-id 20090820112444.998F.52131E4D@oss.ntt.co.jp
Whole thread Raw
In response to Re: FDW-based dblink (WIP)  (Tom Lane <tgl@sss.pgh.pa.us>)
Responses Re: FDW-based dblink (WIP)  (Heikki Linnakangas <heikki.linnakangas@enterprisedb.com>)
List pgsql-hackers
Tom Lane <tgl@sss.pgh.pa.us> wrote:

> I don't believe there is any consensus for integrating dblink into core,
> and I for one will resist that strongly.  Keep it in contrib.

OK, our consensus is that dblink should be replaced with SQL/MED interface
and then we'll start to consider integrating into core.

However, automatic transaction management needs help by core. Is it
acceptable to have two-phase callbacks? Registered callbacks are
called with TWOPHASE_EVENT_PRE_COMMIT when a transaction is about
to be committed or prepared. The argument gxact is NULL if the
transaction is committed without 2PC.

    typedef enum
    {
        TWOPHASE_EVENT_PRE_COMMIT,
        TWOPHASE_EVENT_POST_COMMIT,
        TWOPHASE_EVENT_POST_ABORT,
        TWOPHASE_EVENT_RECOVER,
    } TwoPhaseEvent;

    typedef void (*TwoPhaseEventCallback) (
        TwoPhaseEvent event, GlobalTransaction gxact, void *arg);

Regards,
---
ITAGAKI Takahiro
NTT Open Source Software Center


Attachment

pgsql-hackers by date:

Previous
From: Robert Haas
Date:
Subject: Re: auto_explain log_verbose causes regression failure
Next
From: Itagaki Takahiro
Date:
Subject: Re: auto_explain log_verbose causes regression failure