Re: Global Interface Table Problem.

Tech-Archive recommends: Speed Up your PC by fixing your registry



Hello Mr Igor,

Thanks for the reply and I am sorry for my late reply I live in different
time zone.
In your post below you have said that "The apartment (thread) in which the
COM object was created is dead, and
the object died with it". If that was the case then dont you think it
should have behaved uniformaly in all the threads. As I had mentioned in my
post that when I create GIT and other interfaces in the primary thread of
IE, close this window and move on to the window created by the primary
thread using File -> New or CTRL N. in this second window GIT world
perfectly. But the vise versa is not happenning.

You also asked me "Why not just implement a regular C++ class in your DLL"
.. The answer to that is that there are other modules present which are
called from the hooks and these modules can be in any language but they all
support com. So we thought if we use com to store our data it will be easy
for us to pass it to these modules.


"Igor Tandetnik" <itandetnik@xxxxxxxx> wrote in message
news:uTcWSTceGHA.1792@xxxxxxxxxxxxxxxxxxxxxxx
Dinesh Venugopalan <dinesh@xxxxxxxxxxx> wrote:
The problem I am facing is this.
Lets say we have 2 IE window A and B. Where A is the
primary window and B is create by doing "file" -> "new" on A.

If the pointers are created in A and if I close A and record in B my
GIT queries work properly.
But if I create pointers in B, close the B window and record in A.
GIT query fails and returns the error values as 0x80070057 which
translates to "The parameter is incorrect".

The apartment (thread) in which the COM object was created is dead, and
the object died with it.


Personally, I'd not use a COM object in this scenario at all. Why not just
implement a regular C++ class in your DLL? Or perhaps a COM-like object,
but don't use COM mechanisms for creting and managing it. Just make the
object thread-safe and share it directly in a global variable.
--
With best wishes,
Igor Tandetnik

With sufficient thrust, pigs fly just fine. However, this is not
necessarily a good idea. It is hard to be sure where they are going to
land, and it could be dangerous sitting under them as they fly
overhead. -- RFC 1925




.



Relevant Pages

  • Re: [git pull] x86 changes for v2.6.27
    ... Please pull the latest x86 git tree from: ... There is also a _ton_ of build fixes which could have been easily ... window to squash patches is small in an append-only setup. ...
    (Linux-Kernel)
  • Re: LogFS merge
    ... When sending patches that happens automatically. ... git the same trick. ... I can still see is the timing with the merge window. ...
    (Linux-Kernel)
  • Re: [GIT PULL] x86: use arch/x86/include
    ... git should cope with this if changes come in via git. ... It has been discussed many times to keep arch and arch include files ... the minimum set of patches are queued up would be best. ... next merge window? ...
    (Linux-Kernel)
  • Re: Just where IS Chuck
    ... interest out the window. ... You stopped being "interesting" halfway through your 1st post! ... You were never interesting from the git go. ... and my personal Google profile has me listed as A**hole. ...
    (alt.vacation.las-vegas)
  • Re: Netzero version QS 7.3 changed my default resizing pointers for IE 6.0.28 and I cant restore th
    ... > I installed Netzero version QS 7.3 as a trial a few days ago. ... > size of the IE window to fit beneath the ad bar. ... > The correct resizing pointers show for all other applications. ... > vertical resize pointers that I have specified in Control ...
    (microsoft.public.windows.inetexplorer.ie6.browser)