bug#70934: Document how to debug failing tests

Bruno Haible <[email protected]> Tue, 14 May 2024 13:59:04 +0200
Newsgroups gmane.comp.gnu.libtool.bugs
Message-ID <1851116.QZUTf85G27@nimes>
Hi,

While running the libtool test suite in a continuous integration environment,
I saw two test failures. In order to investigate these test failure, I had
to understand how to pass additional options, and which options, to the
testsuite "program". I did not find this information in the HACKING file.
Therefore here is a patch to add it to the HACKING file.
0001-Document-how-to-debug-failing-tests.patch (text/x-patch, 1.1 KB)
From d9a6da60797c94f7195e801f59416b3978bfdb42 Mon Sep 17 00:00:00 2001
From: Bruno Haible <[email protected]>
Date: Tue, 14 May 2024 13:25:51 +0200
Subject: [PATCH 1/2] Document how to debug failing tests.

* HACKING (Test Suite): Explain how to debug a failing range of tests.
---
 HACKING | 6 ++++++
 1 file changed, 6 insertions(+)

diff --git a/HACKING b/HACKING
index 20d65e9a..c419d953 100644
--- a/HACKING
+++ b/HACKING
@@ -76,6 +76,12 @@ and is not part of a release distribution.
   AT_CHECK) is on a single line so that 'testsuite -x' displays the
   whole thing.  You can use m4_do or '[... ]dnl' to wrap long lines.
 
+* To debug a failing range of tests, e.g. 13 to 14:
+    make check-local TESTSUITEFLAGS="--debug 13-14"
+  More generally, in the TESTSUITEFLAGS you can use the documented
+  options for Autotest generated testsuites:
+  https://www.gnu.org/savannah-checkouts/gnu/autoconf/manual/autoconf-2.72/html_node/testsuite-Invocation.html
+
 * Run tests in parallel with
     make -k check TESTSUITEFLAGS="--jobs=$(nproc)"
   liberally, on as many platforms as you can.  Use as many compilers and
-- 
2.34.1