🚀 go-pugleaf

"Mike Collins" <its@TheBottomOfThePost.com> wrote in message
news:45c145d9$1@newsgroups.borland.com...

> we have now started to have reports, from the US, that the help
> dialog is failing to be displayed and the error message returned
> seems to relate to the CoInitialize call failing.

Please be more specific.  What exactly is happening?  What is the
actual error code being returned?  Is it actually being reported when
you call CoInitialize() yourself, or when the dialog is being created?

>       catch (const Exception &exception)
>       {
>          // ************* This seems to be the error message that is
> displayed **************

Given the code you have shown, the error has to be coming from the
dialog itself, since that is the only place a VCL exception can be
getting thrown from.  It would be helpful to know exactly WHICH line
is throwing the exception, though.

Probably the dialog is making its own call to CoInitialize/Ex()
(directly or indirectly) and then failing because CoInitialize() was
previously called with a different threading model specified.

> The problem seems to be in the initial call to CoInitialize, and the
error
> message that our users are seeing is that displayed from the last
catch
> statement (Exception &exception).

What is the actual error message?

> I know this is a bit general

Too general.

> should i sit in a loop calling CoInitialize until i get a success.

Absolutely not.

> If is possible that the necessary services will not have been fully
> started at that point?

No, because you already one one call to CoInitialize() succeeding
before the error ocurs.


Try this code instead:

    if( SUCCEEDED(CoInitialize(NULL)) )
    {
        try
        {
            THelpOptDlg *HelpOptDlg = NULL;

            try
            {
                HelpOptDlg = new THelpOptDlg(NULL);
            }
            catch (...)
            {
                GinaStub->WlxMsgBox(DISPLAY_HELP_DLG_CREATE_ERROR_MSG,
MB_OK | MB_ICONERROR | MB_TASKMODAL | MB_TOPMOST);
            }

            if( HelpOptDlg )
            {
                try
                {
                    try
                    {
                        iModalRst = HelpOptDlg->ShowModal();
                    }
                    __finally
                    {
                        delete HelpOptDlg;
                    }
                }
                catch (const Exception &)
                {

GinaStub->WlxMsgBox(DISPLAY_HELP_DLG_SHOW_ERROR_MSG, MB_OK |
MB_ICONERROR | MB_TASKMODAL | MB_TOPMOST);
                }
            }
        }
        catch (const Exception &e)
        {
            AnsiString szExpMsg;

szExpMsg.sprintf(GinaStub->RtnSz(KID_SAFE_GENERAL_EXCEPTION_MSG),
e.Message.c_str());
            GinaStub->WlxMsgBox(szExpMsg, MB_OK | MB_ICONERROR |
MB_TASKMODAL | MB_TOPMOST);
        }

        // Must be called once the form with the corrisponding OLE
container is destroyed
        CoUninitialize();
    }
    else
        GinaStub->WlxMsgBox(DISPLAY_HELP_OLE_ERROR_MSG, MB_OK |
MB_ICONWARNING | MB_TASKMODAL | MB_TOPMOST);


Gambit

Remy, thanks a hundred fold, you reply was very useful

I've just given it a quick scan but the points you raised are very valid - 
i'll give it my full attention when i get back.

The problem that i'm having is that i cannot reproduce the error - we have 
done hundreds of sales and thus installations without issue.  It has only 
been reported by a handfull of people.  To add to this problem, all the 
uninstallation funcitonality and live update functionality WAS only 
accessible from the help - so if the dialog fails, as it has, we knacked. 
There is no, proactical was to remove the software.  This was obviously a 
design flaw by myself and has been re-addressed - there are now other ways 
to access the said fucntionality.

But because of this, i can't send out updates to try and test to see if 
anything resolves the problem.  I did strip out the Help dialog and place it 
in a stand alone applicaiton that mimiced exactly, what was going on and 
sent this to our clients, however theis executed without issue and displaye 
dthe dialog correctly.

I have not tracked down a UK customer who is allowing me access to there 
laptop tomorrow so i should be able to provide you with a beter 
explaintaiton and locate the point of failour.

I had thought that it might be possible that some of the OLE / COM 
underlying services might not be fully initialised at such an early point in 
the startup of the system.



"Remy Lebeau (TeamB)" <no.spam@no.spam.com> wrote in message 
news:45c154b4$1@newsgroups.borland.com...
>
> "Mike Collins" <its@TheBottomOfThePost.com> wrote in message
> news:45c145d9$1@newsgroups.borland.com...
>
>> we have now started to have reports, from the US, that the help
>> dialog is failing to be displayed and the error message returned
>> seems to relate to the CoInitialize call failing.
>
> Please be more specific.  What exactly is happening?  What is the
> actual error code being returned?  Is it actually being reported when
> you call CoInitialize() yourself, or when the dialog is being created?
>
>>       catch (const Exception &exception)
>>       {
>>          // ************* This seems to be the error message that is
>> displayed **************
>
> Given the code you have shown, the error has to be coming from the
> dialog itself, since that is the only place a VCL exception can be
> getting thrown from.  It would be helpful to know exactly WHICH line
> is throwing the exception, though.
>
> Probably the dialog is making its own call to CoInitialize/Ex()
> (directly or indirectly) and then failing because CoInitialize() was
> previously called with a different threading model specified.
>
>> The problem seems to be in the initial call to CoInitialize, and the
> error
>> message that our users are seeing is that displayed from the last
> catch
>> statement (Exception &exception).
>
> What is the actual error message?
>
>> I know this is a bit general
>
> Too general.
>
>> should i sit in a loop calling CoInitialize until i get a success.
>
> Absolutely not.
>
>> If is possible that the necessary services will not have been fully
>> started at that point?
>
> No, because you already one one call to CoInitialize() succeeding
> before the error ocurs.
>
>
> Try this code instead:
>
>    if( SUCCEEDED(CoInitialize(NULL)) )
>    {
>        try
>        {
>            THelpOptDlg *HelpOptDlg = NULL;
>
>            try
>            {
>                HelpOptDlg = new THelpOptDlg(NULL);
>            }
>            catch (...)
>            {
>                GinaStub->WlxMsgBox(DISPLAY_HELP_DLG_CREATE_ERROR_MSG,
> MB_OK | MB_ICONERROR | MB_TASKMODAL | MB_TOPMOST);
>            }
>
>            if( HelpOptDlg )
>            {
>                try
>                {
>                    try
>                    {
>                        iModalRst = HelpOptDlg->ShowModal();
>                    }
>                    __finally
>                    {
>                        delete HelpOptDlg;
>                    }
>                }
>                catch (const Exception &)
>                {
>
> GinaStub->WlxMsgBox(DISPLAY_HELP_DLG_SHOW_ERROR_MSG, MB_OK |
> MB_ICONERROR | MB_TASKMODAL | MB_TOPMOST);
>                }
>            }
>        }
>        catch (const Exception &e)
>        {
>            AnsiString szExpMsg;
>
> szExpMsg.sprintf(GinaStub->RtnSz(KID_SAFE_GENERAL_EXCEPTION_MSG),
> e.Message.c_str());
>            GinaStub->WlxMsgBox(szExpMsg, MB_OK | MB_ICONERROR |
> MB_TASKMODAL | MB_TOPMOST);
>        }
>
>        // Must be called once the form with the corrisponding OLE
> container is destroyed
>        CoUninitialize();
>    }
>    else
>        GinaStub->WlxMsgBox(DISPLAY_HELP_OLE_ERROR_MSG, MB_OK |
> MB_ICONWARNING | MB_TASKMODAL | MB_TOPMOST);
>
>
> Gambit
>
> 

Ok, got a bit more information - sitting infront of a system that is 
exibiting the problem not - remember that we have hundreds of installations 
that work fine - so i'm at a lost to the actual cause.  Whilst i can now say 
what the error message is, and thus where it must be stemming from, i've 
know idea why one system should fail and another work

The error message that i get is an EAccessViolation - which is being catcht 
and shown from the final catch statement:

catch (const Exception &exception)

- the title of the error message confirms this also.  So i'm not a million 
miles futher forward - excpet that i know that somewhere between the outer 
most try / catch, a VCL EAccessViolation is being raise..

Any ideas? 

"Mike Collins" <its@TheBottomOfThePost.com> wrote in message
news:45c31da6@newsgroups.borland.com...

> The error message that i get is an EAccessViolation

That is a long way away from a CoInitialize problem.

> which is being catcht and shown from the final catch statement:

Wht does your code actually look like now?  Which catch statement are
you referring to?  Which line is actually throwing the exception to
begin with?


Gambit

Hay Remy, I've found the erroneous code, but I'm going to report a new item 
as it's nothing to do with the CoInitialize


"Remy Lebeau (TeamB)" <no.spam@no.spam.com> wrote in message 
news:45c393cb$1@newsgroups.borland.com...
>