[unixODBC-dev] Re: Symbol name collision between unixODBC and driver

Nick Gorham nick at lurcher.org
Thu Nov 3 19:27:29 GMT 2005


Marc Herbert wrote:

>Marc Herbert <Marc.Herbert at emicnetworks.com> writes:
>
>  
>
>>Do the symbols have the same name so some people can shortcut the
>>driver manager and hardwire a driver with minimum effort?
>>
>>Or was it designed like this just to scare away non- dlopen() experts? :-)
>>    
>>
>
>Or maybe it was the other way round, driver managers being "invented"
>_after_ bare drivers and slipped inbetween...
>  
>
You would have to go back to the history, the goal was to provide a 
standard API for data access, and also to provide for run time binding 
of drivers to applications. In the land of C and Pascal the sensible 
choice is a driver manager to handle the loading and the passing of 
calls from app to driver. C++ and Java both provide alternative ways of 
doing the same, as in dbExpress and JDBC, but in reality they don't add 
anything to the game. Also, it means the DM can be involved in the 
mapping of calls from one version of the API to the next, You get none 
of that with JDBC, if the driver doesn't support what you call, it 
fails, ODBC allows the careful application writer to handle a wide range 
of drivers with a wide range of capabilities.

-- 
Nick
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://mailman.unixodbc.org/pipermail/unixodbc-dev/attachments/20051103/44b0c0d2/attachment.html>


More information about the unixODBC-dev mailing list