From: | "r(dot)takahashi_2(at)fujitsu(dot)com" <r(dot)takahashi_2(at)fujitsu(dot)com> |
---|---|
To: | 'Yugo NAGATA' <nagata(at)sraoss(dot)co(dot)jp> |
Cc: | 'Zhihong Yu' <zyu(at)yugabyte(dot)com>, vignesh C <vignesh21(at)gmail(dot)com>, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Andrew Dunstan <andrew(at)dunslane(dot)net>, Andy Fan <zhihui(dot)fan1213(at)gmail(dot)com>, PostgreSQL Hackers <pgsql-hackers(at)lists(dot)postgresql(dot)org>, Tatsuo Ishii <ishii(at)sraoss(dot)co(dot)jp>, Thomas Munro <thomas(dot)munro(at)gmail(dot)com> |
Subject: | RE: Implementing Incremental View Maintenance |
Date: | 2021-11-24 04:27:13 |
Message-ID: | OS0PR01MB56827E97C60EDBA790AE719782619@OS0PR01MB5682.jpnprd01.prod.outlook.com |
Views: | Raw Message | Whole Thread | Download mbox | Resend email |
Thread: | |
Lists: | pgsql-hackers |
Hi Nagata-san,
Sorry for late reply.
> However, even if we create triggers recursively on the parents or children, we would still
> need more consideration. This is because we will have to convert the format of tuple of
> modified table to the format of the table specified in the view for cases that the parent
> and some children have different format.
>
> I think supporting partitioned tables can be left for the next release.
OK. I understand.
In the v24-patch, creating IVM on partions or partition table is prohibited.
It is OK but it should be documented.
Perhaps, the following statement describe this.
If so, I think the definition of "simple base table" is ambiguous for some users.
+ IMMVs must be based on simple base tables. It's not supported to
+ create them on top of views or materialized views.
> DEPENDENCY_IMMV was added to clear that a certain trigger is related to IMMV.
> We dropped the IVM trigger and its dependencies from IMMV when REFRESH ... WITH NO DATA
> is executed. Without the new deptype, we may accidentally delete a dependency created
> with an intention other than the IVM trigger.
OK. I understand.
> I think it is harder than you expected. When an IMMV is switched to a normal
> materialized view, we needs to drop hidden columns (__ivm_count__ etc.), and in
> the opposite case, we need to create them again. The former (IMMV->IVM) might be
> easer, but for the latter (IVM->IMMV) I wonder we would need to re-create
> IMMV.
OK. I understand.
Regards,
Ryohei Takahashi
From | Date | Subject | |
---|---|---|---|
Next Message | Amit Kapila | 2021-11-24 04:28:06 | Re: parallel vacuum comments |
Previous Message | Michael Paquier | 2021-11-24 04:20:29 | Re: Some RELKIND macro refactoring |