Re: [PATCH 0/3] Fix thread exited messages for remote targets

[Pedro Alves <pedro@palves.net>]

On 2023-08-21 17:10, John Baldwin wrote:

> One question I have is why is the optional metadata specifically a
> ULONGEST exit code rather than just a std::string suffix?  If it were
> a string then windows-nat.c could use something like:
> 
>    std::string info = string_printf ("with code %u", (unsigned) exit_code).
>    delete_thread_with_info (to_del, info, silent);
> 

The idea was that if other targets want to print something, it'll probably
be the thread exit code as well, so keeping the text formatting centralized
ensures any such target prints it in the same way.  For example, the remote
target's "thread exit" stop reply is "wNN", but we're not printing that thread
exit code today.  Remote debugging with Windows gdbserver ideally would want
to print it, I think.  On Linux, we also have access to the lwp's exit code,
though I think that in practice it's always 0.