forum.vdsworld.com

Dr. Dread · Posted: Sun May 30, 2004 11:51 am Post subject: Error loading VDS 3 (and 4)

For some time I've often been having problems loading VDS 3 (same applies to v. 4). I get a program error:

"The instruction at "x (some address)" referenced memory at "x (same address)". Memory could not read.

The program then just bombs out and cannot start. I'm suspecting the Petite compression coz I have the
same problem with another program compressed with Petite.

Anyone got an idea concerning this? A fix, workaround or something coz this really freaks me out.

Greetz
Dr. Dread

PS: OS = WinXP Home + Pro + W2K

FreezingFire · **Admin Team** Joined: 23 Jun 2002 Posts: 3508

Do you have the latest builds of all of the versions? Smile

Dr. Dread · Posted: Sun May 30, 2004 2:03 pm Post subject:

Of course!

Greetz
Dread

CodeScript · Posted: Sun May 30, 2004 2:41 pm Post subject:

It bombs out pvds too.
I believe it has something to do with COM objects in the VDS runtime which are unchanged in VDS 5 and initially - I think the dll tries to load at a diff address - but after some time it fails. - just my guess.
It has never been fixed - till date - and guess it to be unlikely as it may involve quite some time - and as a side effect it may even benefit - by preventing questions on older VDS versions.

Just to spark discussion - Take no offence

jules · Posted: Sun May 30, 2004 5:54 pm Post subject:

I seem to recall when working on VDS 5 that I detected something that had been in there since VDS 3 (and possibly earlier) that potentially could cause an access violation during start-up. I think it may have had to do with the DDE unit, but it was a long time ago and I could be wrong about that. Actually there were a few things I found that made me think "how on earth has nobody spotted that!"

Vic D'Elfant · Posted: Mon May 31, 2004 8:46 am Post subject:

I'm having that problem too (when using VDS5) Sad

Vic

Dr. Dread · Posted: Mon May 31, 2004 9:25 am Post subject:

I still suspect Petite because the other program I mentioned comes with a compressed and an uncompressed
EXE. The compressed one gives me exactly this error while the uncompressed program always
loads without problems...

Greetz
Dread

FreezingFire · **Admin Team** Joined: 23 Jun 2002 Posts: 3508

I thought that VDS 5 wasn't compressed with Petite though this time. Confused

CodeScript · Posted: Mon May 31, 2004 12:24 pm Post subject:

Even I thought/think so

jules · Posted: Mon May 31, 2004 1:24 pm Post subject:

The VDS IDE is not compressed with anything. The run-time DLLs are compressed with UPX.

If it's something to do with OLE, as CodeScript suggests, then it's entirely possible that Microsoft has changed the behavior of some shared DLL so as to disagree with the way Delphi uses it. This would explain why the problem hasn't come to light until now, and why only certain people are seeing it. The VDS IDE doesn't use OLE or ActiveX directly, but the Delphi VCL code that gets called by the IDE code may use it.

CodeScript · Posted: Mon May 31, 2004 2:10 pm Post subject:

Hi Julian

I had mentioned about that problem long time back and if these info help in resolving it:

1. The problem can be traced to the vdsruntime.
2. Almost the same time as IDE bombs out even the stand-alone exes made using older VDS versions like dialog designer etc for example encounters the problem.
3. VDS5 runtime need to be loaded/unloaded several times before the error begins.
4. Inially the error is intermittent but it soon becomes permanent.
5. Once the error starts the older runtime will re-load only after a complete reboot on XP. A logoff and relogin doesn't help.
6. I haven't encountered the problem on Win 98( but I hardly use Win 98 - so not too sure).
7.PVDS is stronger than VDS 3/4 in withstanding the problem.

I think older version of Prakash's VDSIPP (?made in older Delphi version) has the same conflict with VDS 5.x.
This may favour your thinking that it has got something to do with Delphi.
Also something in Global memory etc goes wrong - related to COM objects or otherwise (vds 5 runtime leaves behind something on unload) as the solution is always a complete reboot.