Re: Setting pd_lower in GIN metapage

On Sun, Sep 10, 2017 at 11:52 AM, Michael Paquier
<michael(dot)paquier(at)gmail(dot)com> wrote:
> On Sun, Sep 10, 2017 at 3:15 PM, Amit Kapila <amit(dot)kapila16(at)gmail(dot)com> wrote:
>> On Sat, Sep 9, 2017 at 9:00 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>>> Michael Paquier <michael(dot)paquier(at)gmail(dot)com> writes:
>>>> On Fri, Sep 8, 2017 at 5:24 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
>>>>> In short, this patch needs a significant rewrite, and more analysis than
>>>>> you've done so far on whether there's actually any benefit to be gained.
>>>>> It might not be worth messing with.
>>>
>>>> I did some measurements of the compressibility of the GIN meta page,
>>>> looking at its FPWs with and without wal_compression and you are
>>>> right: there is no direct compressibility effect when setting pd_lower
>>>> on the meta page. However, it seems to me that there is an argument
>>>> still pleading on favor of this patch for wal_consistency_checking.
>>>
>>> I think that would be true if we did both my point 1 and 2, so that
>>> the wal replay functions could trust pd_lower to be sane in all cases.
>>> But really, if you have to touch all the places that write these
>>> metapages, you might as well mark them REGBUF_STANDARD while at it.
>>>
>>>> The same comment ought to be mentioned for btree.
>>>
>>> Yeah, I was wondering if we ought not clean up btree/hash while at it.
>>> At the very least, their existing comments saying that it's inessential
>>> to set pd_lower could use some more detail about why or why not.
>>>
>>
>> +1. I think we can even use REGBUF_STANDARD in the hash for metapage
>> where currently it is not used. I can give a try to write a patch for
>> hash/btree part if you want.
>
> Coordinating efforts here would be nice. If you, Amit K, are taking
> care of a patch for btree and hash
>

I think here we should first agree on what we want to do. Based on
Tom's comment, I was thinking of changing comments in btree/hash part
and additionally for hash indexes, I can see if we can pass
REGBUF_STANDARD for all usages of metapage. I am not sure if we want
similar exercise for btree as well.

--
With Regards,
Amit Kapila.
EnterpriseDB: http://www.enterprisedb.com