>>I'm curious about something and thought you guys would be a good source of feedback.
>>
>>In order to properly ask my question, I need to provide some groundwork... I'll try to be brief....
>>
>>I have a fairly small application that school teachers use to enter testing data. The application is installed on the teacher's computer and the data (which contains all students in the district) is a VFP table (actually several) and is stored on the district server. (I save the path to where the data is located in a local table.)
>>
>>When the teacher starts the application, they select their name (from a drop down box) and enter the password (that they have chosen.) If they log in correctly, the application sets a filter on the main student table so that the teacher can only view students in their classroom. (One of the fields in the student table is Teacher Code, which is unique to the teacher, so I use it to set the filter with.)
>>
>>This software is being used by about 60 school districts across Texas and has worked without any major problems. (The application was written in 1998 and has been upgraded and improved each year.) Most of the districts using it have about 4 or 5 campuses (approx 40-100 teachers, resulting in approx 800-2000 student records), so there's really not alot of data. As teachers use a toolbar (clicking Next, Previous, etc) to view thier students, speed has been reasonable so far.
>>
>>However, I've recently sold the software to a district which has 42 campuses, which means the student table will contain approx 9,000 records.
>>
>>My question is this: Instead of filtering the data according to the teacher code (i.e. SET FILTER TO TCODE = XXX) would the application perform faster if I re-engineered the application to use views instead.
>>
>>If the performance improvement were drastic, then I certainly would rewrite. However, if the performance improvement would be negligible, then I would prefer to keep things the way they are.
>>
>>Disregarding issues concerning locking, etc, I mainly am interested if I can expect a VIEW to be very much faster than issuing a SET FILTER.
>>
>>I appreciate and welcome any input...
>
>Patric,
>View sounds to be better but arguable how much it could accelarate (if ever) in a 9K system. For a quick take off you might however try 'set key to'. It's easier to test. Get a copy and change 'filter' to 'key'.
>Cetin

BTW with 800-2000 recs 'speed is reasonable' - if you mean it's noticable but acceptable then actually you should investigate bottlenecks in app.
Cetin