Re: Row pattern recognition

Hi Tatsuo,

You are right. I checked and found the 0006 patch I applied was wrong
> one.
>
> Sorry for confusion.
>

No worries. Having 31 patches in the series is bound to cause
confusion — that is on me for sending such a large set at once.

t-ishii$ git log --oneline -10|cat
> 6dd0765463a Row pattern recognition patch (typedefs.list).
> c71940f9e21 Row pattern recognition patch for raw parser.
> 322bab79744 Move declarations related to locktags from lock.h to new
> locktag.h
>

Thank you for the log. I can see that your base is the original v46
application, while mine is the cfbot rebase from April 1
(732acf9b7c6).

No, the commit is not in my working tree.
>
> So I guess I should have rebased v46 tree so that the commit is in the
> work tree before applying your patches. Am I correct?
>

Since the root cause was the patch file mix-up, a rebase would not
have been necessary in this case. That said, the related fix that
went into master ("Fix integer overflow in nodeWindowAgg.c") touches
the same file as several of my patches, so it may cause conflicts
when applying v47. Please let me know if anything is difficult to
resolve.

Let me check the crash first. Since apparently the crash was caused by
> my mis operation.
>

Once confirmed, I will send the corrected patches first.

Regarding the README.rpr suggestion from the 0008 review: the
documentation in execRPR.c has dependencies spread across the patch
series, so separating it mid-review would be disruptive. I plan to
split it out as part of the final patch list once all 31 patches have
been reviewed.

I prefer this way.


Good — I will continue folding review feedback into the relevant
existing patches. Patch numbers and subjects will stay stable across
revisions.

Regards,
Henson