On Mon, 2011-07-25 at 16:58 +0200, Murray Cumming wrote:
At the least there can be a test that there is at least one backend.
Also, maybe we can test for the expected backends based on the
configure
options?
> A unit test might be able to check that some well-known backends
(like
> the file backend) is listed, but anything beyond that is difficult.
It
> would have caught this issue, though.
Yes, I think it would be useful.
This patch adds a "make check" test that just checks whether
SyncSource::backendsInfo() is an empty string. Obviously, we would want
to add more specific API than just backendsInfo() to check for specific
backends.
However, I've added an ifdef there because, of course, we should expect
that to be empty when ENABLE_MODULES is not defined (when using
--disable-shared), because the SyncSource code checks for that:
https://meego.gitorious.org/meego-middleware/syncevolution/blobs/dbus-ser...
So I guess this isn't really a test case for the problem that you are
discussing. What "backends" _should_ be available when using
--disable-shared? And is every backend a source backend?
--
murrayc(a)murrayc.com
www.murrayc.com
www.openismus.com