On 27/06/09 23:01, Samuel Thibault wrote:
I guess I am thinking of this in a very different way or a slightly different point. My idea was that we believe we should have the correct version, but due to no actual linking at compile time the two seem so much more unrelated and so possibly separately distributeable that how can you really be sure that what you believe you have is what you want? My thought is one last check of version numbers. If after all our best attempts to get the correct version (through what ever ways should be tried) reveal no version of our desired version then we could raise a suitable exception. I guess I am thinking, ctypes is doing all its stuff at runtime, therefore we should confirm things at runtime.Michael Whapples, le Sat 27 Jun 2009 22:51:26 +0100, a écrit :My thoughts on the API version stuff, can't we just use the output from the lou_version function to determine whether the bindings and library match?And if they don't, do what? How to guess how the proper version is called to manage to load it?
This relies on us guaranteeing we know what version we have before we load it.Just loading the proper version should always work.
Samuel For a description of the software and to download it go to http://www.jjb-software.com
For a description of the software and to download it go to http://www.jjb-software.com