> Why the app does _nothing_ with dllonexit stubbed is a mystery
> to me, but I am never surprised by any stupid thing a windows programmer
> might do.
:-)
Are you sure its dllonexit thats causing the problem? It shouldn't have an
effect until the dll that called it is unloaded...
> Have you got ideas for this, or should I apply my Man from Mars
> approach? I guess we have to make a linked list by dll of these
> things...any idea what start and end are? In a lot of cases, start is 4
> bytes more than end.
I know what it needs to be done, but I'm not sure if the way I want to do it
is valid. The idea is just like atexit/onexit; the function passed in should
be added to a list of calls to be made on caller exit. The difference is, if
a dll is using msvcrt and so is an application, you cant register the atexit
calls in the _aplications_ exit list (the one maintained inside the dll), the
dll has its own list (delimited by the start and end pointer parameters).
This allows a dll to be loaded/unloaded dymanically without screwing up the
application by calling any atexit funcs it has registered.
The implementation should add the function ptr to the list; what im not sure
is whether you need to grow the list or not. my first stab at it didn't and
it crashed horribly. I think you do, but how do you figure out when to grow
it?
Answers on the back of a postcard please ;-)
I can have a look at this after the unicode stuff is done if you don't think
you can get it going yourself, especially if you think it'll get your app
working with the builtin.
Cheers,
Jon
--
"Once you realise you're not _meant_ to fit in, it all makes sense..."
[EMAIL PROTECTED] , [EMAIL PROTECTED]
