Re: additional contrib test suites

On Sat, Sep 16, 2017 at 5:15 AM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> I wrote:
>> Peter Eisentraut <peter(dot)eisentraut(at)2ndquadrant(dot)com> writes:
>>> So, we have one failure for chkpass on OpenBSD, because OpenBSD crypt()
>>> doesn't support the traditional two-character salt format.
>
>>> Option:
>
>>> - Use the resultmap features to make this an expected failure on OpenBSD.
>
>>> - Fix the module to work on OpenBSD. This would involve making a
>>> platform-specific modification to use whatever advanced salt format they
>>> want.
>
>>> - Replace the entire module with something that does not depend on crypt().
>
>> Or (4) drop the module's regression test again.
>
> Noting that mandrill is showing yet a different failure, one that I think
> is inherent to chkpass:
>
> CREATE TABLE test (i int, p chkpass);
> INSERT INTO test VALUES (1, 'hello'), (2, 'goodbye');
> + WARNING: type chkpass has unstable input conversion for "hello"
> + LINE 1: INSERT INTO test VALUES (1, 'hello'), (2, 'goodbye');
> + ^
> + WARNING: type chkpass has unstable input conversion for "goodbye"
> + LINE 1: INSERT INTO test VALUES (1, 'hello'), (2, 'goodbye');
> + ^
>
> I'm starting to think that (4) might be the best avenue. Or we could
> consider
>
> (5) drop contrib/chkpass altogether, on the grounds that it's too badly
> designed, and too obsolete crypto-wise, to be useful or supportable.

crypt() uses the 7 lowest characters, which makes for 7.2e16 values,
so I would be fine with (5), then (4) as the test suite is not
portable.
--
Michael