On Thu, Dec 10, 2009 at 9:51 AM, Timothy Baldridge <[email protected]> wrote: > To play the devil's advocate here, I'd say that duplicating code is a > bad idea. Now all changes to the un-reduced view must also be made the > reduced view.
This is my issue with that solution too. I don't like repeating myself. The reason my code doesn't know what view it is is because I wrote a layer for communicating with CouchDB that's independent of the view being called, in the interest of saving myself time in the long run. So, I have an agnostic method that I want to query an arbitrary view, with arbitrary start and end keys, and give me non-reduced results. > Secondly, I agree with the OP, it seems odd that CouchDB is so >"relaxed" when it comes to document structure, but a simple thing like > silently disregarding an argument will cause it to complain. I tend to like strict error checking, actually. My gripe really isn't that the error checking is strict, but rather that I can't discover how to send a well-formed request that doesn't get the error. Rather, I have to send the request, see whether I get the error, and send a corrected request if I do. That's sloppy. Best Regards, Luke
