Re: [RFA] gdb.base/c*.exp, send_gdb vs. gdb_test

[Michael Snyder <msnyder at vmware dot com>]

Pedro Alves wrote:
> On Saturday 22 May 2010 22:19:30, Michael Snyder wrote:
>
> > Index: call-ar-st.exp
> > ===================================================================
> > RCS file: /cvs/src/src/gdb/testsuite/gdb.base/call-ar-st.exp,v
> > retrieving revision 1.24
> > diff -u -p -r1.24 call-ar-st.exp
> > --- call-ar-st.exp      5 May 2010 18:06:57 -0000       1.24
> > +++ call-ar-st.exp      22 May 2010 21:15:10 -0000
> > @@ -61,24 +61,19 @@ set timeout [expr "$timeout + 60"]
> >  proc set_lang_c {} {
> >      global gdb_prompt
> >  
> > -    send_gdb "set language c\n"
> > -    gdb_expect {
> > -       -re ".*$gdb_prompt $" {}
> > -       timeout { fail "set language c (timeout)" ; return 0; }
> > -    }
> > +    gdb_test_no_output "set language c"
> >  
> > -    send_gdb "show language\n"
> > -    gdb_expect {
> > +    gdb_test_multiple "show language" "set language to \"c\"" {
> >         -re ".* source language is \"c\".*$gdb_prompt $" {
> >             pass "set language to \"c\""
> >             return 1
> >         }
> >         -re ".*$gdb_prompt $" {
> > -           fail "setting language to \"c\""
> > +           fail "set language to \"c\""
> >             return 0
> >         }
> >         timeout {
> > -           fail "can't show language (timeout)"
> > +           fail "(timeout) set language to \"c\""
> >             return 0
> >         }
>
> You'll need to have the pass branch set a variable, and
> do the return outside gdb_test_multiple.  This is because
> gdb_test_multiple catches other error reasons (the whole purpose
> of gdb_test_multiple).  This pattern shows up in several
> other files, I'll skip mentioning those from here on.

OK, I'll fix it in this instance.  The other instances are
never called, so I'll just remove them.

> (I'm not sure whether you've missed the "return 0" on the
> "set" test above: it does seem okay to let the "show" test
> handle it.)
>
> >  #call compute_with_small_structs(20)
> > -send_gdb "print compute_with_small_structs(20)\n"
> > -gdb_expect {
> > -    -re ".*\[0-9\]+ =.*$gdb_prompt $" {
> > -        pass "print compute_with_small_structs(20)"
> > -      }
> > -    -re ".*$gdb_prompt $" { fail "print compute_with_small_structs(20)" }
> > -    timeout           { fail "(timeout) compute_with_small_structs(20)" }
> > -  }
> > +gdb_test "print compute_with_small_structs(20)" \
> > +    "\[0-9\]+ = void" \
> > +    "print compute_with_small_structs(20)"
> >  
>
> You didn't resist changing the expected output here?  :-)

Caught me.

> > -    send_gdb "step\n"
> > -    gdb_expect {
> > -        -re "
> > -init_bit_flags_combo \\(bit_flags_combo=, a=1, b=0, ch1=121 .y., g=1, d=0, ch2=110 .n., e=1, o=0\\) at .*call-ar-st.c:416\[ \t\n\r\]+416.*bit_flags_combo->alpha = a;.*$gdb_prompt $" {
> > -                            pass "step into init_bit_flags_combo"}
> > -        -re ".*$gdb_prompt $" { fail "step into init_bit_flags_combo" }
> > -        timeout { fail "step into init_bit_flags_combo (timeout)" }
> > -    }
> > +gdb_test "step" \
> > +    "init_bit_flags_combo \\(bit_flags_combo=, a=1, b=0, ch1=121 .y., g=1, d=0, ch2=110 .n., e=1, o=0\\) at .*call-ar-st.c:416\[ \t\n\r\]+416.*bit_flags_combo->alpha = a;" \
> > +    "step into init_bit_flags_combo"
> >  
>
> Missing .* at end of expected output?  I saw several of
> those; I don't know if this was on purpose.
>
> On Saturday 22 May 2010 22:19:30, Michael Snyder wrote:
> > +    gdb_test_multiple "bt" "backtrace" {
> > +       -re "#(\[0-9\]*) *<function called from gdb>.*$gdb_prompt $" {
> > +           return $expect_out(1,string)
> > +       }
> > +       -re "$gdb_prompt $" {
> > +           return ""
> > +       }
> > +       timeout {
> > +           return ""
> > +       }
> > +    }
> > +    return ""
>
> Similar remark one of the above, but in this case, there's already a
> return outside gdb_test_multiple, so you can remove the last -re and
> the timeout handling.

OK.

> >  if ![gdb_skip_stdio_test "call str_func1(s)"] {
> > -    send_gdb "call  str_func1(s)\n"
> > -    gdb_expect {
> > -       -re "first string arg is: test string.*\"test string\".*$gdb_prompt $" {
> > -           pass "call str_func1(s)"
> > -       }
> > -       -re ".*$gdb_prompt $" { fail "call str_func1(s)" }
> > -       timeout               { fail "(timeout) call str_func1(s)" }
> > -    }
> > +    gdb_test "call  str_func1(s)" \
>
> The previous test output didn't have the extra space.  Could you
> remove it?  There are several instances of this.

What extra space are you talking about?

> I've been assuming you've been diffing gdb.sum before/after
> these patches; it'd be nice if you mentioned any changes shown by
> that diff.


I run the test after each change, and note any changes in numbers of
pass/fails.


> > +       "first string arg is: test string.*\"test string\".*"
> >  }
>
> Are you planning on committing the patches that have been
> reviewed soon?  I think it would be a good idea to get them in
> sooner than later, so that we can see if there are any problems
> we may have missed in review, and so we'd avoid propagating the
> same mistakes in all these other patches.