Re: Row pattern recognition - Mailing list pgsql-hackers

From Henson Choi
Subject Re: Row pattern recognition
Date
Msg-id CAAAe_zC9pSE5m9ZhqpQ3BDbro30MZVpjvZkJy80Rr=J=5C1Zeg@mail.gmail.com
Whole thread Raw
In response to Re: Row pattern recognition  (Henson Choi <assam258@gmail.com>)
Responses Re: Row pattern recognition
List pgsql-hackers
Hi Ishii-san,

While reviewing the code, I noticed a potential improvement
in 'register_reduced_frame_map()':

diff --git a/src/backend/executor/nodeWindowAgg.c b/src/backend/executor/nodeWindowAgg.c
index 5185ad40237..3a71c279945 100644
--- a/src/backend/executor/nodeWindowAgg.c
+++ b/src/backend/executor/nodeWindowAgg.c
@@ -4770,7 +4770,7 @@ register_reduced_frame_map(WindowAggState *winstate, int64 pos, int val)
        if (pos < 0)
                elog(ERROR, "wrong pos: " INT64_FORMAT, pos);
 
-       if (pos > winstate->alloc_sz - 1)
+       while (pos > winstate->alloc_sz - 1)
        {
                realloc_sz = winstate->alloc_sz * 2;


In practice, 'pos' is always sequential (0, 1, 2, ...),
so the current 'if' works fine - doubling once is always
sufficient for +1 increment.

However, using 'while' would be more defensive, handling
any theoretical case where 'pos' might jump by more than
2x the current size.

It's a trivial change with no functional impact on normal
operation, but makes the code more robust.

What do you think?

Best regards

pgsql-hackers by date:

Previous
From: David Rowley
Date:
Subject: Re: Patch: dumping tables data in multiple chunks in pg_dump
Next
From: Rob Moore
Date:
Subject: Proposal: Add backup start time to pg_stat_progress_basebackup