Hi everybody again, (Resending this EMail again because apparently I have just send in HTML format, which wasn't my intention)
on our production servers I have quite some errors due to excessively long varchar-values which application-code tries to insert into tables and which don't fit. The errors look like ERROR: value too long for type character varying(4) This is not helping me much. The patch will turn this too ERROR: value too long for type character varying(4) (hint: column-name is mycolumn) if the column that was overflown was mycolumn. The tables have many columns, the statements are not readable and many columns happen to have the same length. Powers of 2 most often for some odd reason ... I fired up gdb and saw that the error message is generated during the preprocessing of the query where some kind of the constant-folding/constant-elimination happens on the parse-tree. I went ahead and added a try/catch at some point upwards in the call-stack where at least i have the contact of the T_TargetEntry. That has a field resname which gives me exactly the information i need... The column which was overflown. With that info i can fix the application code much more easily. Relation name was out of reach for me, there is a void* passed transparently to the constant-mutator but that is not checkable at the point. That context contains the original top-level statement node however. The patch just adds a bit of hinting to the error message and goes on.. That is all but really helpful to me and potentially also others. Attached Patch has more Infos and comments. Regards from Germany, Stepan Stepan Rutz Phone: +49 (0) 178 654 9284 Email: stepan.r...@gmx.de Earth: Brunnenallee 25a, 50226 Frechen, Germany
columname_hint.patch
Description: Binary data
-- Sent via pgsql-hackers mailing list (pgsql-hackers@postgresql.org) To make changes to your subscription: http://www.postgresql.org/mailpref/pgsql-hackers
