|
Using WinXP, VS2005, MFC project
We have a number of COM DLLs we register via a batch file that contains a call to regsvr32 for each DLL.
Problems
1) The regsvr32 app seems to be respawning on its own with anywhere from 10-20 instances of itself running at any given time (and instances coming and going, seemingly at random). How do I fix that?
2) I can't seem to get one of our interfaces to register. I've tried to make sure that all of the spellings are the same, and such, but it doesn't seem to want to take (doesn't show up in the registery). Can anyone list things to check?
------- sig starts
"Why don't you tie a kerosene-soaked rag around your ankles so the ants won't climb up and eat your candy ass..." - Dale Earnhardt
"...the staggering layers of obscenity in your statement make it a work of art on so many levels." - Jason Jystad, 10/26/2001
|
|
|
|
|
John Simmons / outlaw programmer wrote: The regsvr32 app seems to be respawning on its own with anywhere from 10-20 instances of itself running at any given time (and instances coming and going, seemingly at random). How do I fix that?
Are all instances started from the batch file and all of them are respawning, or do they appear as banshees from virtually nowhere?
What happens if you don't call RegSvr32 with '/s' and add a "pause" between the calls to RegSvr32 in the batch file?
John Simmons / outlaw programmer wrote: I can't seem to get one of our interfaces to register.
Have you written the registration code by yourself or settled with the default implementation in DllRegisterServer?
Is the missing interface declared inside the scope of the library section in the .idl-file?
Can't say that I have a very good idea of what the problem is, but maybe it's possible to narrow it down to be able to test it...
BTW, is it a problem that has appeared all of a sudden or has this always been the case?
--
Roger
It's suppose to be hard, otherwise anybody could do it!
|
|
|
|
|
Roger Stoltz wrote: John Simmons / outlaw programmer wrote:
The regsvr32 app seems to be respawning on its own with anywhere from 10-20 instances of itself running at any given time (and instances coming and going, seemingly at random). How do I fix that?
Are all instances started from the batch file and all of them are respawning, or do they appear as banshees from virtually nowhere?
What happens if you don't call RegSvr32 with '/s' and add a "pause" between the calls to RegSvr32 in the batch file?
The first 10 or so are being started in the batch file. The rest appear out of nowhere, coming and going at will. I haven't tried the pause thing yet.
The same batch file works just fine on another programmer's box - his code does not yet include my new class, but he does have the library my class is in.
Roger Stoltz wrote: John Simmons / outlaw programmer wrote:
I can't seem to get one of our interfaces to register.
Have you written the registration code by yourself or settled with the default implementation in DllRegisterServer?
Is the missing interface declared inside the scope of the library section in the .idl-file?
Can't say that I have a very good idea of what the problem is, but maybe it's possible to narrow it down to be able to test it...
BTW, is it a problem that has appeared all of a sudden or has this always been the case?
I copy/pasted the rgs file from a class that already exists. I've checked, double-check, and triple-checked the contents of that file to make sure it was correct. I haven't tried to manually apply it to the registry yet.
The interface is defined in the IDL file within the scope of the library section (that was the first thing I checked).
It only happens to the new class I added. There are a number of other classes in the same lib that work fine.
------- sig starts
"Why don't you tie a kerosene-soaked rag around your ankles so the ants won't climb up and eat your candy ass..." - Dale Earnhardt
"...the staggering layers of obscenity in your statement make it a work of art on so many levels." - Jason Jystad, 10/26/2001
|
|
|
|
|
I put a pause between each call to regsvr32, and each one returns the following error message:
GetLastError returns 0x0000007e
I off to google that...
------- sig starts
"Why don't you tie a kerosene-soaked rag around your ankles so the ants won't climb up and eat your candy ass..." - Dale Earnhardt
"...the staggering layers of obscenity in your statement make it a work of art on so many levels." - Jason Jystad, 10/26/2001
|
|
|
|
|
Well, system error 0x7e has the following description: "The specified module could not be found".
This makes me wonder if it could be a missing module... 
I understood from your previous post that you have copied an existing .rgs file.
Are you certain that there aren't any references to other modules that aren't copied to the target machine?
Have you registered the proxy/stub dll if you don't use universal marshalling?
Somehow I'm having a hard time finding the connection between this error and the respawning behaviour...
I suspect it's something else.
Are there any other servers in the lib that try to register another server from DllRegisterServer that could explain the respawning behaviour?
It's suppose to be hard, otherwise anybody could do it!
|
|
|
|
|
Ya know, I bet this might have something to do with the fact that I had vs2005 beta installed, and then uninstalled it in order to install the release version.
------- sig starts
"Why don't you tie a kerosene-soaked rag around your ankles so the ants won't climb up and eat your candy ass..." - Dale Earnhardt
"...the staggering layers of obscenity in your statement make it a work of art on so many levels." - Jason Jystad, 10/26/2001
|
|
|
|
|
Umm, could be...
You know what they say about Bill: he works in mysterious ways...
"MicroSoft" is also a not so well-known fabric softener that softens all '1' so they can be bent into '0' without rebuilding... 
More seriously, what are your plans... Wiping the entire disc and in the end reinstalling the release version of VS2005?
If non of the servers in your lib could be registered I would give that one a try, but not when it's only the new one that causes troubles.
Have you tried debugging while registering the server? What happens?
It's suppose to be hard, otherwise anybody could do it!
|
|
|
|
|
Well, I can't register *anything*, but the other programmer in our group can register even the new DLLs, so it's looking like a complete wipe/reinstall is going to be the result (goddamnit).
------- sig starts
"Why don't you tie a kerosene-soaked rag around your ankles so the ants won't climb up and eat your candy ass..." - Dale Earnhardt
"...the staggering layers of obscenity in your statement make it a work of art on so many levels." - Jason Jystad, 10/26/2001
|
|
|
|
|
Perhaps you've already started the reinstall process, but you could try this first....
You mentioned an .rgs file, which usually means that you need another COM server that ships with ATL, namely the ATL Registrar server.
Have a look in the OLE/COM viewer to see if it's registered.
You should find it under its ProgID, ATL.Registrar. The .dll is atl.dll and can usually be found in the system32 dir.
If the Registrar is not registered, it could explain the error code about a missing module.
Give it a shot and let me know, will you?
--
Roger
It's suppose to be hard, otherwise anybody could do it!
|
|
|
|
|
I haven't started it yet -our network guy wants to do it, but he's on a family emergency leave for the rest of the week.
I'll look for that when I get to work this morning, but...
One of the symptoms of the problem is that when I run regsvr32 with no parameters, it takes almost a minute for it to come back with the expected error message. During that time, an additional instance of regsvr32 are spawned several times (it appears and disappears in the task manager process list).
On everyone else's machine, it comes back almost immediately with the expected error message. I honestly think my machine is just hosed up... :/
------- sig starts
"Why don't you tie a kerosene-soaked rag around your ankles so the ants won't climb up and eat your candy ass..." - Dale Earnhardt
"...the staggering layers of obscenity in your statement make it a work of art on so many levels." - Jason Jystad, 10/26/2001
|
|
|
|
|
John Simmons / outlaw programmer wrote: I honestly think my machine is just hosed up...
Well, I suppose you're the better judge of that...
It's suppose to be hard, otherwise anybody could do it!
|
|
|
|
|
Hi Gurus,
What is the difference between COM and ATLCOM..?
which is more powerful ??
Knock out 'T' from CAN'T
You 'CAN' if you think you 'CAN'
-- modified at 9:00 Thursday 4th May, 2006
|
|
|
|
|
hi,
according to my understanding
COM is a specification to write libraries that can be reused among languages.
ATL is a Wizard or Library to create the components.
there is nothing called ATLCOM.
^-^
@|@
- redCat
|
|
|
|
|
Do you have any idea about steps to be follow to write the COM ..?
Thanks in advance.
Knock out 'T' from CAN'T
You 'CAN' if you think you 'CAN'
|
|
|
|
|
COM is the famous Component Object Model.
ATL is the Active Template Library used for creating COM components, ActiveX controls and so on.
Regarding ATLCOM the only thing that comes to mind is the .dll that contains the ATL implementation, atlcom.dll.
For more information about COM, Michael Dunn's article is an excellent starting point and you'll find it here[^].
Hope this helps
--
Roger
It's suppose to be hard, otherwise anybody could do it!
|
|
|
|
|
Thanks Roger
I got it now
Knock out 'T' from CAN'T
You 'CAN' if you think you 'CAN'
|
|
|
|
|
Hi All,
I'm getting a link time error error LNK2001: unresolved external symbol <clsid>. I'm not able to get what is the cause behind it.
Just do it
|
|
|
|
|
the .lib file is not there in the link path.
or you might have declared a function but forgot to define it.
simple see the function name in the linker error.
^-^
@|@
- redCat
|
|
|
|
|
leanmean wrote: I'm getting a link time error error LNK2001: unresolved external symbol . I'm not able to get what is the cause behind it.
Ummm, an unresolved external perhaps...;)
Seriously, you have to tell us more.
An 'unresolved external' means that you somewhere makes a function call to a declared function, but the definition of the function cannot be found by the linker.
In other words: you may have forgotten to add the source file that contains the function implementation to the project so it won't get compiled.
If this doesn't help...
- is the function a part of an external library or
- have you written the function by yourself
Hope this helps
--
Roger
It's suppose to be hard, otherwise anybody could do it!
|
|
|
|
|
i am using a vc++ Automation client to access the methods of a component.
the Component is local server.
i need to call 3 methods from this component. First function f1, i am able to call,
for the second function f2, i am getting this error in HRESULT.
i am not doing any store operation in that method, and the method should return me an xml.
i tried to use the services using IMyInterface* . it seems working fine.
but if i try to access using IDispatch, i am getting the problem.
Please give me some clues if any of you have came across this problem , just help me.
- thanks
^-^
@|@
- redCat
|
|
|
|
|
hi,
i myself has fixed the problem. i am passing a [out]BSTR* parameter to this method. and it is not initialized. after assigning BSTR a = NULL; it seems working
- bye
^-^
@|@
- redCat
|
|
|
|
|
Hi everybody, last (I hope) problem.
After I released a beta of my preoject, 4 months ago, the customer asked me to change my COM-calling style.
I used Smart Pointers with CreateInstance, and server registration on both sides. The customer does not want the server to be registered on client machine, so I was suggested to still use Smart Pointers, but now with CoCreateInstanceEx and Attach.
Briefly, it was:
<br />
ISwitchBoardPtr pISB;<br />
...<br />
pISB.CreateInstance(__uuidof(SwitchBoard));<br />
<br />
...<br />
...my code...<br />
...<br />
pISB.Release
now, it seems it has to be:
ISwitchBoardPtr pISB;<br />
COSERVERINFO csi = {0, L"10.0.1.9", NULL, 0};<br />
MULTI_QI qi = {&__uuidof(ISwitchBoard), NULL, S_OK};<br />
HRESULT hr = CoCreateInstanceEx(__uuidof(SwitchBoard), NULL, CLSCTX_REMOTE_SERVER, &csi, 1, &qi);<br />
ISwitchBoard *_pISB = static_cast<ISwitchBoard *>(qi.pItf);<br />
pSB.Attach(_pISB);
In this object I expose several UDTs, even big and complex (imagine a structure that has as one of its fields a SafeArray containing another structure), and when I used the first method, everything worked fine and the customer was happy, but when I started to use the second method, NONE of my UDTs is marshaled anymore. I get crashes in client (while the server keeps running fine, logically) when I try to get as [out] or [out, retval] parameters one of my UDTs.
I'm stuck. The only thing I thought about is that qi.pItf is a IUnknown * type, while my objects expose IDispatch interfaces, but IDispatch is derived from IUnknown, as every interface in COM, so it should work, shouldn't it?
Thanks for your kind help again...
|
|
|
|
|
How you are populating the UDTs? as safearray? If not universal marshaler cant marshal them as UDTS are unnown type for them.
Also register thr type library on the client if you didnt do so.
rgds..mil10.
|
|
|
|
|
How can the use of smart pointers cause problems with marshalling?
The call is made exactly the same way to the server and the type library is still the same. Am I missing something here?
Since you're mentioning IDispatch I guess that the interface in question is declared in the IDL-file as either dispinterface or has the oleautomation attribute set.
If this is the case, then Mil10's comments applies: you cannot pass UDTs through an automation interface unless you convert them into SafeArrays of e.g. char.
The universal marshaller can only handle types that can be represented with a VARIANT.
If you are indeed using a custom interface that doesn't derive from IDispatch you have to deal with the marshalling by yourself and also distribute and register the marshalling dll that contains the proxy/stub.
As I understood your execution environment, the server is located on another machine than the client. This means that the proxy/stub has to be registered on both machines, otherwise at least the marshalling won't work.
Hope this helps
--
Roger
It's suppose to be hard, otherwise anybody could do it!
|
|
|
|
|
Exactly, either use universal marshler compatiable types, ie Safearrays with variants and register the universal marshler's proxy, ie the type library file on the cleint.
Or Go for custom marshaling and register the custom proxy stub dll on the cleint. I dont think this second option is feasible. It will be better to go for the first option of universl marshaling.
rgds...mil10
|
|
|
|
|
General 
News 
Suggestion 
Question 
Bug 
Answer 
Joke 
Praise 
Rant 
Admin 
Submit an article or tip
