Just an aside, your point about CONNECTION LIMIT was something that just didn't come to my mind and is probably a good way to test ALTER DATABASE with CONNECTION LIMIT.
Its just that that actually wasn't what I was testing there. That 'CONNECTION LIMIT' test was coupled with CREATE DATABASE because I wanted to test that 'branch' in the CREATE DATABASE function just to ensure that there was a regression test that tests CONNECTION LIMIT specifically with CREATE DATABASE. That's all. A check to confirm whether connection limit restrictions actually got enforced was something I missed, but well, its out of the window for now anyway.
I am generally a bit unsure whether the regression tests you propose aren't a bit too verbose. Does any of the committers have an opinion about this?
My feeling is that they are ok if they aren't slowing down things much.
On 2013-06-26 01:55:53 -0500, Robins Tharakan wrote: > The CREATE DATABASE test itself was checking whether the 'CONNECTION LIMIT' > was working. Removed that as well.
You should be able to test that by setting the connection limit to 1 and then try to connect using \c. The old connection is only dropped after the new one has been successfully performed.
