Re: [OLE #16] Comment out an assert, as we don't yet implement IRemUnknown

[Andreas Mohr]

Hi,

On Tue, Dec 28, 2004 at 03:37:17PM +0000, Mike Hearn wrote:
> Mike Hearn <mh(a)codeweavers.com>
> Comment out an assert, as we don't yet implement IRemUnknown
>
> --- dlls/ole32/stubmanager.c  (revision 9)
> +++ dlls/ole32/stubmanager.c  (local)
> @@ -171,8 +175,10 @@ int stub_manager_unref(struct stub_manag
>          EnterCriticalSection(&m->lock);
>          if (!list_empty(&m->ifstubs))
>          {
> -            ERR("PANIC: Stub manager is being destroyed with outstanding interface stubs\n");
> -            assert( FALSE );
> +            ERR("PANIC: Stub manager %s is being destroyed with outstanding interface stubs\n", wine_dbgstr_longlong(m->oid));
> +            
> +            /* assert( FALSE ); */
> +            /* fixme: this will happen sometimes until we do proper lifecycle management via IRemUnknown */
>          }
>
> /* fixme: the lifecycle of table-weak marshals is not

$ find|xargs grep FIXME|wc -l
  11011
$ find|xargs grep fixme|wc -l
     83

Might be good to keep it that way...

Andreas Mohr

[OLE #16] Comment out an assert, as we don't yet implement IRemUnknown

[Mike Hearn]

On Tue, 28 Dec 2004 17:30:42 +0100, Andreas Mohr wrote:
> $ find|xargs grep FIXME|wc -l
>   11011
> $ find|xargs grep fixme|wc -l
>      83
>
> Might be good to keep it that way...

Hmm? I'm not sure what you're saying, sorry. The lowercase fixmes just
look better inside a comment, IMHO, and there are so many that spamming
the user with debug chatter is silly: we already know that we have to fix
it :)

I guess the ERR could become a FIXME but it's not really a situation that
should be fixed. It's an erroneous situation the code may sometimes get
into because of problems elsewhere. So I think the ERR should stay.

thanks -mike

[OLE #16] Comment out an assert, as we don't yet implement IRemUnknown

[Dimitrie O. Paun]

On Tue, Dec 28, 2004 at 06:21:30PM +0000, Mike Hearn wrote:
> Hmm? I'm not sure what you're saying, sorry. The lowercase fixmes just
> look better inside a comment, IMHO, and there are so many that spamming
> the user with debug chatter is silly: we already know that we have to fix
> it :)

It may look better in a comment, but it's easier to spot and grep for
if it's all uppercase. Also, if it's not often called, I think it's
better if we have it as a FIXME();

-- 
Dimi.

[OLE #16] Comment out an assert, as we don't yet implement IRemUnknown

[Mike Hearn]

On Tue, 2004-12-28 at 13:28 -0500, Dimitrie O. Paun wrote:
> It may look better in a comment, but it's easier to spot and grep for
> if it's all uppercase. Also, if it's not often called, I think it's
> better if we have it as a FIXME();

That's true, though it's not hard to find fixmes in this code. It's
still a bit of building site.

That function is called quite often, an average InstallShield spews out
quite a lot of those errors. 

The fixme comments are there mostly for our (me and rob) benefit right
now, most of them will be [re]moved as we work.

thanks -mike