[RFC,gdb/testsuite] Remove breakpoint command in gdb.base/ui-redirect.exp

  Hi,

I noticed a bit in gdb.base/ui-redirect.exp that sets a breakpoint
command on main, but then due to runto_main, the break point is deleted:
...
(gdb) break main^M
Breakpoint 1 at 0x4004d7: file
/data/gdb_versions/devel/src/gdb/testsuite/gdb.base/start.c, line 34.^M
(gdb) commands^M
Type commands for breakpoint(s) 1, one per line.^M
End with a line saying just "end".^M
>PASS: gdb.base/ui-redirect.exp: commands
print 1^M
>PASS: gdb.base/ui-redirect.exp: print 1
end^M
(gdb) PASS: gdb.base/ui-redirect.exp: end
delete breakpoints^M
Delete all breakpoints? (y or n) y^M
...

In the original commit ( submission here
https://sourceware.org/ml/gdb-patches/2010-09/msg00120.html ) there was
no runto_main, and the breakpoint command was not triggered either.

Is this some artefact, and can it be removed, or is it actually testing
something related to redirection?

Thanks,
- Tom
  

* Tom de Vries <tdevries@suse.de> [2019-09-03 19:43:44 +0200]:

> Hi,
> 
> I noticed a bit in gdb.base/ui-redirect.exp that sets a breakpoint
> command on main, but then due to runto_main, the break point is deleted:
> ...
> (gdb) break main^M
> Breakpoint 1 at 0x4004d7: file
> /data/gdb_versions/devel/src/gdb/testsuite/gdb.base/start.c, line 34.^M
> (gdb) commands^M
> Type commands for breakpoint(s) 1, one per line.^M
> End with a line saying just "end".^M
> >PASS: gdb.base/ui-redirect.exp: commands
> print 1^M
> >PASS: gdb.base/ui-redirect.exp: print 1
> end^M
> (gdb) PASS: gdb.base/ui-redirect.exp: end
> delete breakpoints^M
> Delete all breakpoints? (y or n) y^M
> ...
> 
> In the original commit ( submission here
> https://sourceware.org/ml/gdb-patches/2010-09/msg00120.html ) there was
> no runto_main, and the breakpoint command was not triggered either.
> 
> Is this some artefact, and can it be removed, or is it actually testing
> something related to redirection?

I can't comment on the original patch intention, but given this part
of the tests never really did anything useful, and the "commands"
mechanism is tested in gdb.base/commands.exp, I'd be OK with what you
propose below.

I'd suggest give it a few days for anyone else to chip in, then just
commit this as an obvious cleanup.

Thanks,
Andrew

> 
> Thanks,
> - Tom

> diff --git a/gdb/testsuite/gdb.base/ui-redirect.exp b/gdb/testsuite/gdb.base/ui-redirect.exp
> index 4507ac51a2..e48ca72e91 100644
> --- a/gdb/testsuite/gdb.base/ui-redirect.exp
> +++ b/gdb/testsuite/gdb.base/ui-redirect.exp
> @@ -23,23 +23,6 @@ if { [prepare_for_testing "failed to prepare" ui-redirect start.c] } {
>      return -1
>  }
>  
> -gdb_breakpoint main
> -
> -set test "commands"
> -gdb_test_multiple $test $test {
> -    -re "End with a line saying just \"end\"\\.\r\n>$" {
> -	pass $test
> -    }
> -}
> -
> -set test "print 1"
> -gdb_test_multiple $test $test {
> -    -re "\r\n>$" {
> -	pass $test
> -    }
> -}
> -gdb_test_no_output "end"
> -
>  if ![runto_main] {
>      fail "can't run to main"
>      return -1
  

On Tue, 03 Sep 2019 19:43:44 +0200, Tom de Vries wrote:
> I noticed a bit in gdb.base/ui-redirect.exp that sets a breakpoint
> command on main, but then due to runto_main, the break point is deleted:
...
> In the original commit ( submission here
> https://sourceware.org/ml/gdb-patches/2010-09/msg00120.html ) there was
> no runto_main, and the breakpoint command was not triggered either.
> 
> Is this some artefact, and can it be removed, or is it actually testing
> something related to redirection?

It looks to me the patch which added the "runto_main" part:

commit ca1285d17534cff3041c07ac7841288e1b3ba19c
Author: Alan Hayward <alan.hayward@arm.com>
Date:   Fri May 17 14:15:01 2019 +0100
    Add debug redirect option

disabled the regression testing.

"runto_main" should be removed otherwise the testcase does not test anything
(or it just tests less).


Jan
  

On Wed, 04 Sep 2019 12:59:52 +0200, Andrew Burgess wrote:
> I can't comment on the original patch intention, but given this part
> of the tests never really did anything useful, and the "commands"
> mechanism is tested in gdb.base/commands.exp, I'd be OK with what you
> propose below.

The "commands" mechanism is not tested there for "save breakpoints" with
various types of output redirection.


Jan