Re: [PATCH] Fix gdb.cp/no-dmgl-verbose.exp test

[Pedro Alves <pedro@palves.net>]

On 2022-04-29 16:48, Carl Love via Gdb-patches wrote:
> On Fri, 2022-04-29 at 09:14 +0000, Lancelot SIX wrote:

> OK, I added the new test.  It works as expected on PowerPC and on my
> Intel system.
> 
> (gdb) PASS: gdb.cp/no-dmgl-verbose.exp: info types ^std::string$
> break f(std::__cxx11::basic_string<char, std::char_traits<char>, std::allocator<char> >)
> 
> Breakpoint 1 at 0x10: file /home/carll/GDB/build-current/gdb/testsuite/../../../binutils-gdb-current/gdb/testsuite/gdb.cp/no-dmgl-verbose.cc, line 23.
> 
> The additional test is good to have.
> 
> The above suggested changes were tested on both PowerPC and Intel and
> pass as expected.
> 
> Please let me know if the updated patch looks ok.  Thanks.

So the file is called gdb.cp/no-dmgl-verbose.exp.  "no-dmgl" most certainly means "no demangle".

However, the intro comment says:

# Test loading symbols from unrelocated C++ object files.

and indeed, the testcase only compiles an object:

 if { [gdb_compile "${srcdir}/${subdir}/${srcfile}" "${binfile}.o" object {c++ debug}] != "" } {
      untested "failed to compile"
      return -1
 }

... and loads that into gdb:

 clean_restart ${testfile}.o

How is that related to "no demangle verbose" ?  A mystery.

OK.  Now, if we're not going to be testing that setting a breakpoint via the
typedef, like "b f(std::string)", WORKS (I thought GDB had code to make that work,
why doesn't it?), then what's the point of the test?  Testing that the breakpoint
set at the non-typedef expansion of std::string is really no different from setting
a breakpoint at any other C++ function, so we could eliminate the std::string dependency.

It seems like something is missing here.

What did the original submission of the patch that added the test say?