Re: [patch] new testcase: Internal error on unqualified name re-set (PR 11657)

[Pedro Alves <pedro at codesourcery dot com>]

On Wednesday 02 June 2010 15:32:33, Jan Kratochvil wrote:

> +# Create and source the file that provides information about the compiler
> +# used to compile the test case.
> +if [get_compiler_info ${binfile} "c++"] {
> +    return -1
> +}

Is this bit actually needed in this test?

> +
> +if { [gdb_compile_shlib $srcdir/$subdir/$libsrcfile $sofile {debug c++}] != ""
> +     || [gdb_compile $srcdir/$subdir/$srcfile $binfile executable [list debug "c++" shlib=${sofile}]] != ""} {
> +    untested ${testfile}.exp
> +    return -1
> +}
> +
> +clean_restart $executable
> +gdb_load_shlibs ${sofile}
> +
> +gdb_test_no_output "set breakpoint pending yes"
> +gdb_test "break C::C" {Breakpoint [0-9]+ \(C::C\) pending\.} "break C::C"
> +gdb_test_no_output {set variable $brk = $bpnum}
> +
> +gdb_breakpoint "main"
> +gdb_run_cmd
> +gdb_test "" "" "start"

Doesn't runto_main work in this case?

> +
> +set test "breakpoint resolved"
> +gdb_test_multiple {info breakpoints $brk} $test {
> +    -re "<PENDING>" {
> +	fail $test
> +    }
> +    -re "\r\n$gdb_prompt $" {
> +	# FIXME: Check more specific resolution state.
> +	pass $test
> +    }
> +}
> 

So will the test hit an internal error, causing a new FAIL?
How about setting up a kfail with the PR number?

-- 
Pedro Alves