On 28.09.25 05:02, Chao Li wrote:
On Fri, Sep 26, 2025 at 11:00 PM Tom Lane <[email protected] <mailto:[email protected]>> wrote:Yeah. In particular, probably 99% of such Datum arrays also have an associated array of bool isnull flags. IMO it makes exactly zero sense to const-ify the Datums without similar protection for their isnull flags.Based on Tom's comment, I have made the scope a little broader. If a function's "Datum *" parameter is changed to const, then if it has a pairing "bool *isnull" parameter, I make it const as well. Also, if the function has other pointer parameters that can be const, I change them to const as well. See v3 attached.
I have committed this. Note that the documentation for the SPI functions needed to be updated.
As discussed in the other thread about ItemPointer etc., I removed from your patch changes that would have broken externally usable callback function signatures. But that was only a small portion of your patch.
