Re: Tid scan improvements

From: Andres Freund <andres(at)anarazel(dot)de>
To: Edmund Horner <ejrh00(at)gmail(dot)com>
Cc: David Rowley <david(dot)rowley(at)2ndquadrant(dot)com>, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us>, Tomas Vondra <tomas(dot)vondra(at)2ndquadrant(dot)com>, PostgreSQL Developers <pgsql-hackers(at)lists(dot)postgresql(dot)org>
Subject: Re: Tid scan improvements
Date: 2019-03-25 21:48:35
Message-ID: 20190325214835.i6bpqwlhhvprka7d@alap3.anarazel.de
Views: Raw Message | Whole Thread | Download mbox | Resend email
Thread:
Lists: pgsql-hackers

Hi,

On 2019-03-15 18:42:40 +1300, Edmund Horner wrote:
> I've had to adapt it to use the table scan API. I've got it compiling
> and passing tests, but I'm uneasy about some things that still use the
> heapam API.
>
> 1. I call heap_setscanlimits as I'm not sure there is a tableam
> equivalent.

There used to be, but it wasn't clear that it was useful. In core pg the
only caller are index range scans, and those are - in a later patch in
the series - moved into the AM as well, as they need to deal with things
like HOT.

> 2. I'm not sure whether non-heap tableam implementations can also be
> supported by my TID Range Scan: we need to be able to set the scan
> limits. There may not be any other implementations yet, but when
> there are, how do we stop the planner using a TID Range Scan for
> non-heap relations?

I've not yet looked through your code, but if required we'd probably
need to add a new tableam callback. It'd be marked optional, and the
planner could just check for its presence. A later part of the pluggable
storage series does that for bitmap scans, perhaps it's worth looking at
that?

> 3. When fetching tuples, I see that nodeSeqscan.c uses
> table_scan_getnextslot, which saves dealing with HeapTuples. But
> nodeTidrangescan wants to do some checking of the block and offset
> before returning the slot. So I have it using heap_getnext and
> ExecStoreBufferHeapTuple. Apart from being heapam-specific, it's just
> not as clean as the new API calls.

Yea, that's not ok. Note that, since yesterday, nodeTidscan doesn't
call heap_fetch() anymore (there's still a heap dependency, but that's
just for heap_get_latest_tid(), which I'll move into execMain or such).

> Ideally, we can get to to support general tableam implementations
> rather than using heapam-specific calls. Any advice on how to do
> this?

Not yet - could you perhaps look at the bitmap scan patch in the tableam
queue, and see if that gives you inspiration?

- Andres

In response to

Browse pgsql-hackers by date

  From Date Subject
Next Message Andres Freund 2019-03-25 21:52:11 Re: Tid scan improvements
Previous Message Andrew Dunstan 2019-03-25 21:32:42 Re: MSVC Build support with visual studio 2019