Horton

Re: Infinite Interval

Ashutosh Bapat <ashutosh.bapat.oss@gmail.com>

View thread View on postgr.es
From: Ashutosh Bapat <ashutosh.bapat.oss@gmail.com>
To: jian he <jian.universality@gmail.com>, Dean Rasheed <dean.a.rasheed@gmail.com>
Cc: Joseph Koshakow <koshy44@gmail.com>, Tom Lane <tgl@sss.pgh.pa.us>, "Gregory Stark (as CFM)" <stark.cfm@gmail.com>, PostgreSQL Hackers <pgsql-hackers@lists.postgresql.org>
Date: 2023-09-22T07:48:51Z
Lists: pgsql-hackers

Attachments

On Thu, Sep 21, 2023 at 7:21 PM Ashutosh Bapat
<ashutosh.bapat.oss@gmail.com> wrote:
>
> One thing I didn't understand though is the use of
> makeIntervalAggState() in interval_avg_deserialize(). In all other
> deserialization functions like numeric_avg_deserialize() we create the
> Agg State in CurrentMemoryContext but makeIntervalAggState() creates
> it in aggcontext. And it works. We could change the code to allocate
> agg state in aggcontext. Not a big change. But I did not find any
> explanation as to why we use CurrentMemoryContext in other places.
> Dean, do you have any idea?

Following code in ExecInterpExpr makes it clear that the
deserialization function is be executed in per tuple memory context.
Whereas the aggregate's context is different from this context and may
lives longer that the context in which deserialization is expected to
happen.

/* evaluate aggregate deserialization function (non-strict portion) */
EEO_CASE(EEOP_AGG_DESERIALIZE)
{
FunctionCallInfo fcinfo = op->d.agg_deserialize.fcinfo_data;
AggState *aggstate = castNode(AggState, state->parent);
MemoryContext oldContext;

/*
* We run the deserialization functions in per-input-tuple memory
* context.
*/
oldContext = MemoryContextSwitchTo(aggstate->tmpcontext->ecxt_per_tuple_memory);
fcinfo->isnull = false;
*op->resvalue = FunctionCallInvoke(fcinfo);
*op->resnull = fcinfo->isnull;
MemoryContextSwitchTo(oldContext);

Hence I have changed interval_avg_deserialize() in 0007 to use
CurrentMemoryContext instead of aggcontext. Rest of the patches are
same as previous set.

-- 
Best Wishes,
Ashutosh Bapat

Commits

  1. Support +/- infinity in the interval data type.

  2. Avoid integer overflow hazard in interval_time().

  3. Guard against overflow in make_interval().

  4. Fix minmax-multi on infinite date/timestamp values

  5. Optimize various aggregate deserialization functions, take 2

  6. Remove dead code in DecodeInterval()

  7. Accept "+infinity" in date and timestamp[tz] input.

  8. Fix overflow hazards in interval input and output conversions.