[PATCH 8/8] [gdb/dap] Fix race between dap exit and gdb exit

[Tom de Vries]

On 2/20/24 16:51, Tom Tromey wrote:
>>>>>> "Tom" == Tom de Vries <tdevries@suse.de> writes:
> 
> Tom> +@in_gdb_thread
> Tom> +def _on_gdb_exiting(event):
> Tom> +    dap_thread_join()
> 
> I think it's better not to export dap_thread_join and then import it
> here, but instead just add to the event listener exactly where it's
> needed.
> 

Done in a v2.

> Tom> +    global _dap_thread_for_gdb_thread
> Tom> +    _dap_thread_for_gdb_thread = start_thread("DAP", really_start_dap)
>   
> You don't even need a new global if you add the listener here.
> 

OK, reverted back to using _dap_thread in a v2.

> Tom> +def dap_thread_join():
> Tom> +    global _dap_thread_for_gdb_thread
> Tom> +    if _dap_thread_for_gdb_thread != None:
> 
> None of this will be necessary with the above, but FYI in Python it's
> idiomatic to do this comparison with "if blah is not None"

When running test-case gdb.dap/type_check.exp, the dap module is 
imported, but no server is started and consequently there's no dap 
thread.  If we don't test for none here, we get a python exception 
telling us that None doesn't support join.

In the v2, I've rewritten this into an early exit with a comment to make 
it more clear.

Thanks,
- Tom