Re: SSI 2PC coverage

On Wed, Aug 24, 2011 at 9:11 PM, Tom Lane <tgl(at)sss(dot)pgh(dot)pa(dot)us> wrote:
> Alvaro Herrera <alvherre(at)commandprompt(dot)com> writes:
>> After having to play with this, I didn't like it very much, because
>> regression.diffs gets spammed with the (rather massive and completely
>> useless) diff in that test.  For the xml tests, rather than ignoring it
>> fail on an installation without libxml, we use an alternative output.
>
>> Unless there are objections, I will commit the alternative file proposed
>> by Dan.
>
> +1 ... "ignore" is a pretty ugly hack here.
>
> Eventually we need some way of detecting that specific tests should be
> skipped because they're irrelevant to the current system configuration.
> contrib/sepgsql is already doing something of the sort, but it's rather
> crude ...

I'm fairly unhappy with the fact that we don't have a better way of
deciding whether we should even *build* sepgsql. The --with-selinux
flag basically doesn't do anything except enable the compile of that
contrib module, and that's kind of a lame use of a configure flag.

Not that I have a better idea...

--
Robert Haas
EnterpriseDB: http://www.enterprisedb.com
The Enterprise PostgreSQL Company