Re: Row pattern recognition

Hi hackers,

cfbot has been failing to build CF 4460 since it rebased v50 onto
current master. The failure is in readfuncs.c, and it is not
something the posted series did wrong.

9673a0aa92f threaded a ReadNodeContext through the node read path:
pg_strtok() takes it as its first argument now, and so does every
hand-written _readXxx(). _readRPRPattern() was written before that
and still uses the one-argument form, so the file no longer compiles.

The commits involved:

9673a0aa92f Make stringToNode() infrastructure thread-safe --
Michael Paquier, 2026-08-24, the mainline change

db2d99323f1 master of 2026-08-25, the first base under CF 4460 to
carry it

f17fac1c234 "[CF 4460] v50 - Implement row pattern recognition
feature" -- the cfbot branch's current commit, the
posted v50 series applied to db2d99323f1

v50 as posted is cut against master of 2026-08-01, where pg_strtok()
still takes one argument, so nothing is wrong with the series as it
was sent. Only a base carrying 9673a0aa92f needs this, and
f17fac1c234 is what I cut the patch against.

The patch does that one adaptation and nothing else. Only the read
path changed upstream, so nothing else about this node needed
touching.

The attachment carries the number 1003, in the naming the earlier
postings use.

This one goes on top of the increment series already posted, as an
addition the base made necessary rather than anything that series
lacked. More may follow the same way as the base picks up further
mainline changes.

Best regards,
Henson