J.M. The WHERE CURRENT OF cursor syntax has always been rather flaky. For that reason i NEVER use it. I make sure all tables have a primary key, prefereably an autonumber integer. I always do my updates based on that primary key. This also has the benefit that the tables will work with ODBC, if you ever need that functionallity.
Dennis --- ELOEN <[EMAIL PROTECTED]> wrote: > J.M > > you update the field that you qualify your rows (cerreur) > Is the message an error or warning one? > > If warning then it is right > If error then I don't know :-) > > Polychronis T. Kontos > Athens, Greece > > > -----Original Message----- > > From: [EMAIL PROTECTED] [mailto:[EMAIL PROTECTED] Behalf Of > > J.M. GRATIAS > > Sent: Wednesday, June 04, 2003 2:25 PM > > To: RBASE-L Mailing List > > Subject: [RBASE-L] - Issue with CURSOR in RBW6.5++ > > > > > > > > Hi all, > > > > > > Is there any rules that prevent from using severals UPDATE on the > same row > > in a DECLARE CURSOR/WHILE LOOP ? > > > > I am supprise that, in the following command file, the fisrt UPDATE > works > > correctly, but the second one return error 2378 > > 'Cursor is not positionned on a valid row'. > > > > Any idea ? > > > > Using W98SE and RBW 6.5++ build 1.851xRT03 ... > > > > > > DECLARE c2 CURSOR FOR SELECT numcde FROM T2commande WHERE cerreur > IS NULL > > OPEN c2 > > > > WHILE #PI <> 0 THEN > > FETCH c2 INTO vnumcde > > IF SQLCODE = 100 THEN > > BREAK > > ENDIF > > ......... > > UPDATE T2commande SET cerreur = 99 WHERE CURRENT OF C2 > > ........ > > SET RULES ON > > UPDATE T2commande SET cerreur = NULL WHERE CURRENT OF C2 > > ...... > > ENDWHILE > > > > J.M. GRATIAS, Logimatique > > >
