Re: [PATCH v2] gdb: make "start" breakpoint inferior-specific

[Simon Marchi <simark@simark.ca>]

On 11/10/22 12:47, Pedro Alves wrote:
> On 2022-11-10 5:33 p.m., Simon Marchi wrote:
> 
>>>> +proc do_test {} {
>>>> +    # With remote, to be able to start an inferior while another one is
>>>> +    # running, we need to use the non-stop variant of the protocol.
>>>> +    save_vars { ::GDBFLAGS } {
>>>> +	if { [target_info gdb_protocol] == "extended-remote"} {
>>>> +	    append ::GDBFLAGS " -ex \"maintenance set target-non-stop on\""
>>>> +	}
>>>> +
>>>> +	clean_restart ${::binfile_other}
>>>> +    }
>>>> +
>>>> +    gdb_test "run&" "Starting program: .*" "start background inferior"
>>>
>>> I was going to point out that if the inferior prints something, this can
>>> timeout, as that output would appear after the prompt.  I then looked around
>>> the tree for "run&" uses, to confirm we are using gdb_test_multiple with that,
>>> and found that you just recently added "gdb_test -no-prompt-anchor", for exactly
>>> this scenario.  :-)  I think that should be used here.
>>
>> Even if, in this case, we know the inferior won't print anything?
> 
> Admitedly it's a bit pedantic, but it seems to me to be safer.  Say
> someones adds some logging to the program or something.  It just looks
> like good practice to me to not have an anchor when the inferior is left
> running after the prompt is printed.

Ack, I will add it.

Simon