Bill Schwab wrote:
> > > > Have you tried that random example at all? > > Yes, and it worked - not much help, I know. Blair had suggested it as an > intermediate at one point to rule out problems. > > Have a good one, > > Bill > Can you tell me how you ran it? If I do a clean install of Dolphin, should I be able to double click on RandomVBClient.exe and have it work? Does the Dolphin installation program take care of registration for the component or do I have to enter some commands in a workspace first and then save the image (or run regsvr32 on Dolphin or anything else)? The answer is probably different depending on whether or not Dolphin is already running when RandomVBClient executes. It probably doesn't work if there are old entries in the registry either. Keith |
Keith,
> Can you tell me how you ran it? If I do a clean install of Dolphin, should I be able to > double click on RandomVBClient.exe and have it work? Does the Dolphin installation > program take care of registration for the component or do I have to enter some commands > in a workspace first and then save the image (or run regsvr32 on Dolphin or anything > else)? It was definitely not installed as part of the setup at the time; Blair had me download something and manually register a type library. So far, I'm finding type libraries to be very poor at self-registering, so I probably used Dolphin's component wizard to load/register the library. For my own components, I've started installing the type libraries to the system directory, and having the corresponding server load/register from there each time it runs. The random stream samples does not appear to be part of my current D5 installation, but, I might be missing something. Have a good one, Bill -- Wilhelm K. Schwab, Ph.D. [hidden email] |
In reply to this post by Keith Alcock-3
OK, it is finally working. In order to use the Dolphin development system (not a deployed app) as a
local server without it already running when the client executes, then 1) The image must be called dolphin.img. 2) The image must be located in the same directory as the client. Maybe these requirements are tighter than necessary, but if you do this, there is a chance that it might work; don't, and you may get very little indication of why it doesn't and it will take a month to figure it out. |
"Keith Alcock" <[hidden email]> wrote in message
news:[hidden email]... > OK, it is finally working. In order to use the Dolphin development system (not a deployed app) as a > local server without it already running when the client executes, then > > 1) The image must be called dolphin.img. > 2) The image must be located in the same directory as the client. > > Maybe these requirements are tighter than necessary, but if you do this, there is a chance that it > might work; don't, and you may get very little indication of why it doesn't and it will take a month > to figure it out. I explained this in my reply of 12 June (yes, almost a month ago) to your posting of 11 June in the thread "Any docs for creating DLLs or COM with D5?". To summarize: The registry entries created for Local Servers by the #register method are only correct for a deployed application, not a development image. When the LocalServer32 keys is mapping to "dolphin.exe" the value must also include the full path to the image, otherwise it will try to load "dolphin.img" from whatever is the working directory (which may or may not be the same directory as the client). I included a patch that modifies the method so that it also registers a development image correctly, depending on whether it is run from a deployed app or a development session. I have reproduced the latest version of this patch below. With this patch you can also request a registration of just in-proc or out-of-proc servers by passing the appropriate CLSCTX values to the #register: method, therefore to register the random stream sample for out-of-process debugging you would evaluate: COMRandomStream register: (OLEConstants at: #CLSCTX_LOCAL_SERVER) Regards Blair --------------------------------- |
"Blair McGlashan" <[hidden email]> wrote in
news:ag3pmk$ic9sm$[hidden email]: > I explained this in my reply of 12 June (yes, almost a > month ago) And you've just replied to an almost 1 month old message! (The message to which that was a reply was posted on June 11, according to my server) P. |
Paul Hudson wrote:
> > "Blair McGlashan" <[hidden email]> wrote in > news:ag3pmk$ic9sm$[hidden email]: > > > I explained this in my reply of 12 June (yes, almost a > > month ago) > > And you've just replied to an almost 1 month old message! > > (The message to which that was a reply was posted on June 11, > according to my server) > > P. There does appear to be a discontinuity in the space-time continuum. Thanks, Blair, for looking into this again. I haven't had time to complete the experiments with COM, so I will no doubt pest you with further questions. IIRC I had finished 3 of the 4 combinations of run-time vs. development session manager and in-proc vs. out-of-proc component. The fourth wasn't behaving yet, but it will have to wait. Thanks. Keith |
In reply to this post by Paul Hudson
"Paul Hudson" <[hidden email]> wrote in message
news:Xns9242C57B2BE40phudsonpoboxcom@127.0.0.1... > "Blair McGlashan" <[hidden email]> wrote in > news:ag3pmk$ic9sm$[hidden email]: > > > I explained this in my reply of 12 June (yes, almost a > > month ago) > > And you've just replied to an almost 1 month old message! > > (The message to which that was a reply was posted on June 11, > according to my server) Yes, sorry about that. I have no idea why it popped up as a "new" message in my news browser. Perhaps it was waiting for a DCOM timeout :-). Regards Blair |
Free forum by Nabble | Edit this page |