You should not do this: orderby=mytablename+'.'+myfield+' '+myorder+','+mytablename+'.id'
You should instead use one of the following orderby=db.mytablename.myfield|db.mytablename.id orderby=~db.mytablename.myfield|db.mytablename.id orderby=db[mytablename][myfield]|db[mytablename].id orderby=~db[mytablename][myfield]|db[mytablename].id This may be vulnerable to SQL Injection orderby=mytablename+'.'+myfield+' '+myorder+','+mytablename+'.id' On Sunday, 21 October 2012 12:52:38 UTC-5, apinho wrote: > > Ok, > > found the answer. By adding 'id' as an orderby field, the order of the > rows if never changed, even if the original orderby field is not unique : > > rows = db().select( > mytable.ALL, > orderby=mytablename+'.'+myfield+' '+myorder+','+mytablename+ > '.id', > limitby=(page*ITEMSPERPAGE-ITEMSPERPAGE,page*ITEMSPERPAGE) > ) > > On Sunday, October 21, 2012 6:19:16 PM UTC+1, apinho wrote: >> >> Hi, >> >> I'm using this query : >> rows = db().select( >> mytable.ALL, >> orderby=mytablename+'.'+myfield+' '+myorder, >> limitby=(page*ITEMSPERPAGE-ITEMSPERPAGE,page*ITEMSPERPAGE) >> ) >> 'myorder' is either 'asc' or 'desc. 'mytable', 'myorderby' and 'myorder' >> ar all string, that I get through request.vars... >> >> The DB is Postgresql. >> >> This works well when 'myfield' is unique within the DB. But when >> 'myfield' has repeated values, running this same query, with the same >> parameters doesn't always return the same set of rows. >> >> So the question is : When the orderby field is not enough to fully order >> tables, cause the field has repeated values, what order does Web2py apply ? >> Or, maybe, it's Postgre related, and not web2py ? >> >> thx >> >> >> >> >> > --
