From: Riggs, R. <RR...@do...> - 2002-03-28 22:03:34
|
MS says "In addition to working directly with these file types, Jet can work against any data source which provides an ODBC driver. This is called attaching an external ODBC data source or table." http://support.microsoft.com/default.aspx?scid=kb;EN-US;q154643 Orbit-python is not necessarily ustable, but, when I used it, O-P was not robust enough for non-trivial applications. I had issues with memory leaks, and with threading issues. And I ran into enough problems with the overridden import function when my dependency tree became even a little complex that I was forced to give up on it. Orbit-python was a great learning tool for me. But I quickly ran into it and Orbit's limitations. Setting up the omniORB name service is fairly trivial. I created omniORB RPMs that includes omniNames. And once the ref is in omniNames, you can reference it with a simple "corbaname" URI. It takes 5-10 lines of Python code to add the reference to the server. Then fetch it with: o = orb.string_to_object("corbaname::myserver#myobject") Lots easier than making your own custom IOR daemon IMHO. -----Original Message----- From: dman [mailto:dm...@dm...] Except that the commercial program we have that uses the database hooks directly into the Jet engine. AFAIK that can't be redirected to an ODBC link. (that would be the preferred option, though) ... Did you find orbit-python to be unstable on linux? The biggest difference I found, so far, between orbit-python and omniORBpy is that the latter requires generating python source from the idl. I also need a way for the client to find the object. I found out that I can't just serve the IOR through a socket and have the orb still work. Setting up a nameserver seems a bit of overkill for this application, but I'll work through the docs on it. I did use a nameserver once, with java, in a class at school. Hmm, perhaps I can exec a separate daemon for serving the IOR. |