Thread: Re: Document how to run only specific regress tests

> On 29 Oct 2024, at 05:40, Paul A Jungwirth <pj@illuminatedcomputing.com> wrote:
>
> I never knew this was possible, but I ran across a note on the meson
> wiki page[1]. I wish I knew about it years ago!
>
> This patch adds a note to
> https://www.postgresql.org/docs/current/regress-run.html so that other
> hackers and potential hackers will know about it.

While not related to this patch per se, the lack of any meson related
information on that page isn't great.

+    Note that sometimes tests depend on each other, so you might get
+    extra failures.

Reading this pretending I don't know about regression test internals, I think
"extra" would be hard to interpret the meaning of.  How about "unexpected"?

--
Daniel Gustafsson

Daniel Gustafsson <daniel@yesql.se> writes:
> While not related to this patch per se, the lack of any meson related
> information on that page isn't great.

Indeed.  I think this is one of the main unfinished bits of business
for our meson conversion.  It's not clear to me that our meson
infrastructure can do all the same testing tricks that the makefiles
can, and for sure it's not documented how.

            regards, tom lane

hi.

maybe verbose, the following is the my intended changes based on your patch.
what do you think?


  <sect2 id="regress-subset">
   <title>Running Specific Tests</title>
   <para>
  You can run a subset of the regression tests through the <envar>TESTS</envar>
  environment variable.
  One way to do it is set <envar>TESTS</envar> environment variable beforehand:
<screen>
<userinput>export TESTS="test_setup boolean char"</userinput>
</screen>
    You can also specify it while invoking the regress tests command,
one of example:
<screen>
make check-tests TESTS="test_setup boolean char"
</screen>
    If the <envar>TESTS</envar> environment variable is set when the regression
    tests are run, it overrides the original <envar>TESTS</envar> environment
    variable.
</para>
<para>
    All the regression tests are stored in <filename
    class='directory'>src/test/regress/sql</filename>.  Each
individual regression test
    name is the corresponding file name omit the file name extension
    (<literal>.sql</literal>).  For example, the regression test name of
    <literal>src/test/regress/sql/partition_join.sql</literal> is
    <literal>partition_join</literal>.  Tests specified by the
<envar>TESTS</envar>
    environment variable will be run in a sequential order.
    <literal>TESTS="test_setup copy' </literal> will first run
    <literal>test_setup</literal> then <literal>copy</literal>.  Note that
    sometimes one test may depend on other tests, so you might get unexpected
    failures if the dependent test is not specified.  For example, to run
    <literal>select_parallel</literal> test, to avoid failure, you can use
<screen>
make check-tests TESTS="test_setup create_misc create_index select_parallel"
</screen>
   </para>
  </sect2>