Re: debugger startup

[Alexandre Julliard]

eric pouech  writes:
> either we decide that shell scripts shouldn't be used (easy
> solution) or we could do something like :
> - if the CreateProcess fails, then wait for a certain amount of
>
> time. If after that time the current process isn't currently
> debugged, then report the error. Otherwise assume what did happen is
> what has been described above

CreateProcess already waits a couple of seconds for the new process to
show up, so it's probably enough to simply check for the debugger once
CreateProcess returns.

An easier fix may be to have the script exec the debugger instead of
launching it as a child process; this way the debugger parent pid will
be the original Wine process, and everything should work fine.

debugger startup

[eric pouech]

> An easier fix may be to have the script exec the debugger instead of
> launching it as a child process; this way the debugger parent pid will
> be the original Wine process, and everything should work fine.

well, this has a major holes:
- since the process creation in Wine knows the created process is not
a PE image, it doesn't set the info on the server side for the newly
created process ; after the exec in the shell script, the info cannot
be retrieved from the parent, hence an error

furthermore, the event used to synchronize the debugger startup with 
the exception handler no longer works (mainly because the inheritage is 
not done thru the preceding step

perhaps, the best fix would be to merge the init_process/new_process
server calls

A+