>>>>>>Hi,
>>>>>>
>>>>>>Would it be correct to assume that when SQL Select from VFP against SQL Server fails and the error is
>>>>>>
>>>>>>

>>>>>>Communication link failure
>>>>>>

>>>>>>
>>>>>>that the problem was that the SQL Connection timed out by the time user executed the SQL Select? Or if not, what else would be causing this type of error?
>>>>>>
>>>>>>TIA
>>>>>
>>>>>No, when the timeout is occurred then the error will be:
>>>>>"Timeout occurred", something like this.
>>>>>
>>>>>Communication link failure usually means that you do not have connection with SQL Server anymore.
>>>>>Some LAN troubles or if the SQL Server is on WAN maybe Internet connection is lost?
>>>>
>>>>A follow up question, please. Do you think if - on determining that this is the error message - I add a code to try to establish the connection to the SQL Server again?
>>>
>>>Usually before we issue any SQL command we try something that SHOULD return result.
>>>(That is very well stolen from Andy Kramek and Marcia Akins)
>>>

>>>IF SQLEXEC(handler goes here, "SELECT @@IDENTITY","crsTest") < 0
>>>   ** What? we have no connection
>>>   ** do what ever you wish. Quit the Application
>>>   ** try to re-establish connection
>>>ENDIF
>>>

>>
>>I have some places in my program where I do almost exactly the same; checking the connection. I was just concerned that if I do it as you, BEFORE ANY SQL command, I will slow down the application. But knowing that you are an expert in SQL Server and probably tested the effect of this approach on speed of the program. So I will follow your example.
>>Thank you!
>
>In the general case, your test code could succeed, but your payload code could still fail. So your payload still has to be fault tolerant - check error return codes or be wrapped in TRY...CATCH etc., and react appropriately. Why not just wrap your payload with the same fault tolerance you need to use for the test code?
>
>Long ago I used to use pre-tests for functions with some likelihood of failure but if one has well structured error handling and a general defensive-programming philosophy I no longer see the need.

Thank you.