Subject Re: locate works differently 2.8 and 9.5 From Mervyn Bick <invalid@invalid.invalid> Date Sat, 24 Jan 2015 16:10:10 +0200 Newsgroups dbase.getting-started On Sat, 24 Jan 2015 12:48:55 +0200, Antti Tanskanen   <antti.tanskanen@elisanet.fi> wrote: > Hi > > I have dbase versions 2.8 8.x and 9.5. I have found that many things   > work differently in these version and maybe there are still some bugs in   > the calculations for running same data gives different results on   > different versions. > > Now I discovered that locate don't work the same way: 2.8. works   > correctly and 9.5 seeks all the parameter length (len_) down to 1. The   > idea behind this code is to find a parameter from atcluo file where the   > vnrker is > 0 and the atc code is longest possible (most accurate). > Any idea why 9.5 (and 8.x) do not work like 2.8 (and as i want). There are MANY ways in which commands in dBASE 8 and dBASE 9 work   differently to the same commands in dBASE 2.8.  Some of these are   intentional, most of them are bugs.  You don't say in what way LOCATE in   dBASE 9.5.1 performs differently to LOCATE in dBASE 2.8 but you should   report this in the bug-report newsgroups together with a little program   that shows the problem. Try the following code which doesn't use the LOCATE command.  It loops   through the table looking for a matching record and then uses the GOTO   command to leave the record pointer on the record. nLen = 7 nRec = 0 select atcluo for n = nLen to 1 step -1     cSeek = left(vnro-len_->atc,n)     scan        if cSeek = left(atc,nLen) or (cSeek = trim(left(atc,nLen)) and   vrnker > 0)           n = 1   //exit for...next           nRec = recno()           exit    //exit scan...endscan        endif     endscan next if nRec = 0     msgbox("No match found") else     goto nRec endif Mervyn.