Attached is a patch that adds the requested support. This is in relation
to the thread at

http://www.gossamer-threads.com/lists/python/python/584264

In addition to the two methods I also "fixed" Fetch() by ensuring that
when the records are exhausted None is returned instead of an error.

If something further is required of me or I submitted this patch
incorrectly, please let me know so I don't screw it up next time. :-)

This patch appears to be mostly duplicated by patch 2125 which has been
accepted. Would it be helpful if I separated out the parts that have now
been accepted?

Yes, please do provide an updated patch.

With apologies for the delay, I have modified the patch to remove the
stuff that has been added already. Some of the other changes are to make
use of new C API functionality but they can be ignored, if desired. The
changes that are relevant are in the view_fetch() routine. Again, if you
have questions, let me know.

Martin, would this go in a bugfix release or is it a new feature for 3.3 only?

I lost track what problem precisely this patch solves, so it's hard to tell whether it's a bug fix or not.

All, I have trimmed this patch down to the bare minimum required to solve this problem. Please review this as I would dearly love to have this committed. The error received without this patch is

MSIError: unknown error 103

This occurs when there are no more records to fetch from the view. The patch simply returns None when there are no more records to fetch -- similar to what is done by the DB API.

Can someone review the latest patch please as it's only five additional lines of C code.

I note that this patch has still not been accepted! Please let me know what needs to be done. I just tried against Python 3.4 and the patch works as expected -- other than the starting line number is now 715. I have signed the contributor agreement if that is what is holding things up.

Sorry, I have no familiarity with msilib, _msi, or the internals of MSIs in general.  The patch looks reasonably harmless to me, but I don't have the confidence to take responsibility for it.

I've opened PR 4459 to fix this. Here's a simple reproducer I adapted from the script uploaded by uday kiran in issue 32064.

New changeset bdb8315c21825487b54852ff0511fb4881ea2181 by Berker Peksag in branch 'master':
bpo-1102: View.Fetch() now returns None when it's exhausted (GH-4459)
https://github.com/python/cpython/commit/bdb8315c21825487b54852ff0511fb4881ea2181

Will it be added to 3.6 also?
And when it will be available in the windows installations you release?

> Will it be added to 3.6 also?

Yes, I've created a backport PR: PR 4520

> And when it will be available in the windows installations you release?

The dates for 3.6.4 release is listed at https://www.python.org/dev/peps/pep-0494/#id4

New changeset 4b3042900e7e8dc120408bab86642c09c9d25a5a by Berker Peksag in branch '3.6':
bpo-1102: View.Fetch() now returns None when it's exhausted (GH-4459)
https://github.com/python/cpython/commit/4b3042900e7e8dc120408bab86642c09c9d25a5a

Thank you, Anthony. I added a test case and committed your patch to 3.6 and master branches.