gdb-7.8

[Fernando de Oliveira]

​http://ftp.gnu.org/gnu/gdb/gdb-7.8.tar.xz

[Fernando de Oliveira]

There are many sources of errors for the test suite, most well known.

I will rephrase a little what is in the book, but not much.

Readline 6.x is a source of errors:

gdb.base/completion.exp: completion of field in anonymous union
https://sourceware.org/ml/gdb-patches/2011-05/msg00002.html
https://build.opensuse.org/package/view_file?file=gdb-bz701131-readline62-2of3.patch&package=gdb&project=devel%3Agcc&rev=faca1ad22a8a7c1d46e3790a16f62e29

The suggested workaround (in fact the readline-5.1 code state) going to patch
into the proposed FSF GDB bundled readline-6.2.  Distros then can stop using
--with-system-readline to still feature recent and system matching readline
until readline-7.0 gets released.

Also some comments at
https://lists.gnu.org/archive/html/bug-readline/2011-04/msg00012.html

Dropping --with-system-readline switch, compared results are:

                            system  readline    bundled readline
# of expected passes        27710   96,32%      27837   96,44%
# of unexpected failures    737     2,56%       709     2,46%
# of unexpected successes   1       0,00%       1       0,00%
# of expected failures      55      0,19%       55      0,19%
# of known failures         37      0,13%       38      0,13%
# of unresolved testcases   24      0,08%       21      0,07%
# of untested testcases     48      0,17%       48      0,17%
# of unsupported tests      156     0,54%       156     0,54%
Total                       28768               28865

I will refer to original using --with-system-readline switch, in the following.

FAIL for timeout    581        78,83%
FAIL other causes   156        21,17%
Total               737

Still more curious are same tests failing for more than one reason (this is what I interpret, may be wrong interpretation).

Failed more than 10 times:

          Test                    Number of failures
           wp-replication.exp     94
              py-inferior.exp     79
                    store.exp     57
           py-prettyprint.exp     42
                schedlock.exp     36
     hand-call-in-threads.exp     33
                      tls.exp     32
                  structs.exp     30
                  restore.exp     30
             multi-create.exp     30
         watch_thread_num.exp     25
                  inux-dp.exp     23
                 pthreads.exp     20
          watchpoint-fork.exp     16
               complaints.exp     15
           siginfo-thread.exp     14
               watchpoint.exp     12
                 selftest.exp     11
     no-unwaited-for-left.exp     11
 local-watch-wrong-thread.exp     11
                        TOTAL    621

I spent some time since yesterday or the day before (can't remember) searching for ther errors and found with success many previous reports.

[Fernando de Oliveira]

Fixed at r13760.