1 Revision history for Test-Valgrind
3 1.19 2016-08-01 21:00 UTC
4 + Fix : valgrind 3.1x will no longer be treated as valgrind 3.1.0,
5 causing the wrong command line arguments to be used.
6 Thanks Lucas Nussbaum and Gregor Herrmann from Debian for
7 reporting and providing feedback (Debian bug 832833).
8 + Fix : The number of callers is now capped at 24, as this is the
9 maximum number supported by valgrind. An higher number of
10 frames could lead to the generation of unusable suppressions
11 in both old and recent versions of valgrind.
13 1.18 2015-11-16 15:00 UTC
14 + Chg : A new Test::Valgrind::Version class has been added to represent
15 valgrind version numbers, instead of lazily relying on
17 + Fix : The detection of the valgrind executable has been slightly
18 improved to cover some edge cases.
19 + Tst : Test failures on Windows, or with old versions of Test::More or
20 File::Temp, have been addressed.
21 Thanks Paul Howarth for reporting.
22 + Tst : A few extraneous warnings displayed by some tests when they
23 were run with old versions of Test::Harnes were silenced.
25 1.17 2015-11-13 13:50 UTC
26 + Fix : [RT #108873] : Tests fail due to "Text file busy"
27 Thanks Paul Howarth for reporting.
29 1.16 2015-11-12 23:30 UTC
30 + Chg : The Test::Valgrind tests will now be skipped when the default
31 and user-supplied suppressions files do not refer to any perl-
32 related symbol. This behaviour can be overridden by passing
33 'allow_no_supp => 1' to Test::Valgrind->import.
34 This fixes the following RT issue.
35 + Fix : [RT #101934] : t/20-bad.t failing on armv7hl
36 While the root cause of this issue is probably not at
37 Test::Valgrind's level, it should nevertheless not run the
38 tests when the suppression files are obviously insufficient.
39 Thanks Paul Howarth for reporting and providing helpful
41 + Fix : The accuracy of the default perl suppression file has been
43 + Fix : The tests will be more reliably skipped when no valgrind or
44 no suppressions are found.
45 + Fix : Segmentation faults during the analysis are now more gracefully
47 + Fix : 'no_def_supp => 1' will no longer cause the extra suppressions
49 + Tst : t/20-bad.t will no longer run the extra tests when no valgrind
50 can be found (this was a regression in version 1.15).
52 1.15 2015-10-30 16:15 UTC
53 + Add : The new 'regen_def_supp' option can be passed to
54 Test::Valgrind->import to forcefully regenerate the default
56 + Fix : Build failures of the dummy XS code with PERL_IMPLICIT_SYS
58 + Fix : Handshake failures in tests with recent perls built with
60 + Tst : Improved diagnostics on failure.
61 + Upd : Freshen Makefile.PL.
63 1.14 2013-09-01 17:10 UTC
64 This is a maintenance release. The code contains no functional change.
65 Satisfied users of version 1.13 can skip this update.
66 + Doc : POD headings are now properly linkable.
67 + Tst : Author tests are no longer bundled with this distribution.
68 They are only made available to authors in the git repository.
69 + Tst : The stack traces used in t/20-bad.t have been made more
70 predictable when compiler optimizations are in use.
72 1.13 2011-08-25 15:45 UTC
73 + Fix : The session will now call the finish() method at the right time.
74 + Fix : The README file is now properly UTF-8-encoded.
75 + Tst : Test demangling of Z-encoded symbols.
77 1.12 2009-11-29 16:45 UTC
78 + Doc : The front documentation in Test::Valgrind was improved.
79 + Fix : Don't install Valgrind.so, as it's only used for testing.
80 + Fix : "Non-portable hex number" warnings.
81 + Fix : Old versions of valgrind could leave some Z-encoded symbols in
82 the stack traces. We now make sure they're always decoded.
84 1.11 2009-10-08 20:55 UTC
85 + Doc : Completion and improvements.
86 + Fix : Work around Kwalitee test misfailures.
87 + Upd : Freshen Makefile.PL.
89 1.10 2009-09-22 18:45 UTC
90 + Chg : valgrind is no longer required to be installed before installing
92 + Chg : The parser logic that used to be located inside the tool was
93 moved to a new Test::Valgrind::Parser hierarchy. The tool
94 decides which parser should be used by implementing the
95 "parser_class" method.
96 + Fix : Compatibility with valgrind 3.5.
97 + Fix : The "valgrind" option to Test::Valgrind::Session->new wasn't
99 + Fix : The "Test" action could generate inappropriate error messages.
101 1.02 2009-05-02 12:05 UTC
102 + Add : Commands can now be aggregated.
103 + Add : The commands can now also filter and mangle reports.
104 + Chg : The perl suppressions are now stripped from everything below
106 + Doc : Typos, nits and clarifications.
108 1.01 2009-04-14 21:15 UTC
109 + Add : Allow testing code given by -e. Hurray for source filters!
110 + Fix : Lazily load version.pm in Test::Valgrind::Session so that it's
111 not really needed at configure time.
112 + Fix : Don't unload dynamic extensions by default so that their symbols
113 still appear in the stack traces.
115 1.00 2009-04-12 22:50 UTC
116 Complete rewrite. The options passed to Test::Valgrind->import have
117 changed, so please have a look at the doc.
118 + Add : A brand new reusable API.
119 + Add : Suppressions are now versionized following the perl interpreter
120 and valgrind. They are regenerated as needed and stored in the
122 + Add : memcheck output is parsed by XML::Twig.
123 + Add : The output of the original script is now correctly captured.
124 + Rem : Lots of ugly hacks in the toolchain.
126 0.08 2009-02-08 18:25 UTC
127 + Add : Unload the libraries loaded by DynaLoader at END time, reducing
128 the number of leaks in perl.
129 + Rem : The useless t/21-good-block.t.
130 + Upd : META.yml spec updated to 1.4.
132 0.07 2008-09-30 13:35 UTC
133 + Chg : Now conforms to the Test::Builder framework.
134 + Fix : STDERR is longer eaten.
135 + Fix : Eat STDOUT or print it as diagnostics, but never let him go
136 through naked, as this may fiddle with the tests output.
137 + Fix : 'make test' without prior 'make'.
139 0.06 2008-09-09 22:10 UTC
140 + Fix : Suppressions were installed in the wrong path. Thanks Marvin
141 Humphrey for reporting this.
142 + Fix : Really load XSLoader when generating suppressions. Thanks Marvin
143 Humphrey again for bringing this to my attention.
144 + Fix : The logic for finding the file to run through valgrind was
145 flawed, which lead to generate suppressions against
146 Test/Valgrind.pm only.
148 0.051 2008-09-02 17:20 UTC
149 + Fix : Forgot to upgrade version of the suppression placeholder module.
151 0.05 2008-08-25 22:05 UTC
152 + Chg : Lower the default caller to 12.
153 + Tst : Add a real-life test that gets executed only if you have a C
156 0.04 2008-05-13 16:25 UTC
157 + Chg : Don't smoke this on 5.005 please.
158 + Chg : Better way to find the valgrind executable.
160 0.03 2008-04-21 15:35 UTC
161 + Fix : Include the mock Suppressions.pm rather than an empty file.
163 0.02 2008-04-21 15:25 UTC
164 + Add : Test::Valgrind now depends on Perl::Destruct::Level. This is
165 needed for non-debugging perls because we can't set their
166 level of memory cleanup correctness on exit with the
167 PERL_DESTRUCT_LEVEL environment variable.
168 + Add : Hardcode valgrind path into the new constant
169 Test::Valgrind::Suppressions::VG_PATH.
170 + Chg : Test::Valgrind::Suppressions::supppath() is now supp_path().
171 + Chg : lib/Test/Valgrind/Suppressions.pm.tpl was renamed to
172 lib/Test/Valgrind/Suppressions.tpl for file portability reasons.
174 0.01 2008-04-19 15:50 UTC
175 First version, released on an unsuspecting world.