Re: msctf: Always return FALSE in DllCanUnloadNow

[Sebastian Lackner]

On 02.04.2015 03:34, Alistair Leslie-Hughes wrote:
> Hi,
> Fixes https://bugs.winehq.org/show_bug.cgi?id=31579
>
> Changelog:
>       msctf: Always return FALSE in DllCanUnloadNow
>
> Best Regards
>   Alistair Leslie-Hughes

I doubt that the patch has any real disadvantage, but wouldn't it be better to fix the refcounting instead of just removing this feature? Qian Hong even attached a suggested patch on the bugreport.

msctf: Always return FALSE in DllCanUnloadNow

[Alistair Leslie-Hughes]

Hi Sebastian,

The patch on the bug only fixes a crash when using the ITfThreadMgr 
interface.

If you created any of the other interfaces in the dll, there would be
chance it would still crash since the DLL could be unloaded since 
MSCTF_refCount would be zero.

My patch is consistent with other dlls, i.e. msxml3, mshtml, atl,
to ensue that the DLL isn't unloaded while an interface may exist.

Best Regards

Alistair Leslie-Hughes
> Date: Thu, 2 Apr 2015 03:37:48 +0200
> From: [e-mail]
> To: [e-mail]
> CC: [e-mail]
> Subject: Re: msctf: Always return FALSE in DllCanUnloadNow
>
> On 02.04.2015 03:34, Alistair Leslie-Hughes wrote:
> > Hi,
> > Fixes https://bugs.winehq.org/show_bug.cgi?id=31579
> >
> > Changelog:
> >       msctf: Always return FALSE in DllCanUnloadNow
> >
> > Best Regards
> >   Alistair Leslie-Hughes
>
> I doubt that the patch has any real disadvantage, but wouldn't it be better to fix the refcounting instead of just removing this feature? Qian Hong even attached a suggested patch on the bugreport.

msctf: Always return FALSE in DllCanUnloadNow

[Qian Hong]

Hi Alistair,

Thanks for working on this bug!
> If you created any of the other interfaces in the dll, there would be
> chance it would still crash since the DLL could be unloaded since
> MSCTF_refCount would be zero.

Right, I tried to test more interface and tried to implement
MSCTF_refCount based on information from all related interfaces
refcounts, but it turns out too complicate and then I gave up.

On Thu, Apr 2, 2015 at 2:51 PM, Alistair Leslie-Hughes
 wrote:
> My patch is consistent with other dlls, i.e. msxml3, mshtml, atl,
> to ensue that the DLL isn't unloaded while an interface may exist.

Seems a good solution to me, sadly I didn't know these examples, if I
know I'd prefer this solution as well.