Re: [PATCH] deferred trace

[Alexandre Julliard]

eric pouech  writes:
> using the Wine debugger (or any program using the debugging API) may
> have some drawbacks. As the API is designed, it's not possible to
> get control of the debugged program between its creation and its
> entry point entering this makes very difficult to let the Wine
> debugger trigger the expected features by poking the memory in the
> debuggee.

That's not exactly correct, the debugger can intercept the dll loads
and break in the dll init code. Since with dll separation all Wine
inits will be done inside dlls, the debugger should be able to
intercept everything from the point we connect to the wineserver.

[PATCH] deferred trace

[eric pouech]

Alexandre Julliard wrote:
> eric pouech  writes:
> > using the Wine debugger (or any program using the debugging API) may
> > have some drawbacks. As the API is designed, it's not possible to
> > get control of the debugged program between its creation and its
> > entry point entering this makes very difficult to let the Wine
> > debugger trigger the expected features by poking the memory in the
> > debuggee.
>
> That's not exactly correct, the debugger can intercept the dll loads
> and break in the dll init code. Since with dll separation all Wine
> inits will be done inside dlls, the debugger should be able to
> intercept everything from the point we connect to the wineserver.

I thought that the process creation event was sent much later in the
process
init code (after DLLs had been loaded). It's no longer the case (for
almost 
one year :-/).

A+