> On Mar 6, 2019, at 6:32 AM, Ron <ronljohnsonjr@xxxxxxxxx> wrote: > > On 3/6/19 1:45 AM, Rob Sargent wrote: > [snip] >> This construct had been working until recent changes but I cannot relate the message to any deformity in the current schema or code. >> Any pointers appreciated. > > What were the recent changes? > > -- > Angular momentum makes the world go 'round. > diff optimalMonoPed.sql optimalMonoPed.sql.~20359ea9e67ddf009db89b94140f67988862f247~ 13,14d12 < imkr int; < jmkr int; 53c51 < -- --- > -- 67,68d64 < select imkr=min(ordinal), jmkr=max(ordinal) from mrkidx where ordinal between segp.firstmarker and segp.lastmarker; < raise notice 'seg % start=% i=% end=% j=%', segp.id, segp.firstmarker, imkr, segp.lastmarker, jmkr; In other words two variables dropped along with the select which set them for their only use in a NOTICE, plus white space on a comment line. So yes the problem must be in the caller. A not on line numbers: Using \ef on this function presents a slight variation of my code: it rearranges the “language plpgsql” from after the final END; (old style I guess) to before the AS. So line 65 is actually what I thought was line 64. Still not the real problem of course. (I’ll update my ways re: coding functions)
