PbDLL bug ... ?

[Lance Edmonds]

Great stuff! Can you please send it along? Thanks!


------------------
Lance
PowerBASIC Support
mailto:[email protected][email protected]</A>

OK. I have a zipfile as per BEFORE fixing, data directory included.

------------------

[Lance Edmonds]

Ok, I've passed this to R&D... they may still request the code, so please put a copy of the faulty code aside if possible. Thanks!

------------------
Lance
PowerBASIC Support
mailto:[email protected][email protected]</A>

No longer necessary to send the code, except the snippet below -
I have done some R&D, and resolved the situation. Invoice follows
soon.

This has solved the gpf problem. The 'Case "Budget amendments"' was
added AFTER the End Select (show below as a remark), and there was no
End Select at the bottom; it has now been added in and the upper one
removed. That is the fix, and once one finds a bug, all becomes clear,
rather like mud, plus or minus!

Case "Quotations in progress checklist"
   Call RemoveMainMenu
   Call RemoveSubMenu
   ECopt&=7
   JCopt&=13
   ActiveMenu$="Job costing"
   ScrnName$="Quotes in progress checklist"
   Call MainHeading
   ViaLbx&=1
   Call CheckQuo (pDlg&)
'   End Select           

  Case "Budget amendments"
   Call RemoveMainMenu
   Call RemoveSubMenu
   ECopt&=7
   JCopt&=16
   ActiveMenu$="Job costing"
   ScrnName$="Budget amendments"
   Call MainHeading
   ViaLbx&=1
   Call BudgeWip (pDlg&)
   End Select

Thanks for your trouble, anyway (I note that 1Mb is as nothing,
for future reference!).


------------------

[Lance Edmonds]

~1Mb? No problem.

Just zip it up and send it over... basically, we need to be able to duplicate the problem in order to get it fixed!

Thanks!!!

------------------
Lance
PowerBASIC Support
mailto:[email protected][email protected]</A>

Lance, thanks. Yes, the error remains consistent, also after reboot.

BUT, if a compilable runnable set of files is needed ... the .Bas and
.Inc etc. files zip up to 1,091,950 bytes. Is that not too big to send
along? (Let me guess: Oh dear, yes, that's MUCH too big!)

Please advise!

Paul, the code I ADDED is a big module, included into an existing
DLL module after testing was completed as a separate .EXE file.
I do this because in the initial .EXE version I have the debug
facility and it is in any event also quicker to just run, rather
than to compile the DLL, call up the main .EXE, run to test, then
recall the DLL file, modify, recompile, call up the EXE ... ad
infinitum. I have been working this way for some time now, and it
really saves time and frustration, and has given no problems
until this time.



------------------

[Paul Dwyer]

See how much of the code you can get rid of and still cause the error (assuming that it persists), then post it so we can try in on different systems to be sure.

We haven't seen an update in a year or so (sorry lance) so we'll want to know what to avoid if it is actually a problem

Cheers

------------------
Paul Dwyer
Network Engineer
Aussie in Tokyo

[Lance Edmonds]

Yes, it could be a bug or it could be that your O/S was unstable due to a prior GPF, or a buggy driver, or a faulty RAM chip, or... the list is endless!

Essentially, if the compile-time GPF is repeatable (even after a reboot), then you'll need to send your code to Tech Support. Once we can verify there is a problem, we'll pass it to R&D to implement a fix.

Thanks!


------------------
Lance
PowerBASIC Support
mailto:[email protected][email protected]</A>

After some amendments/additions to code, I compiled the module -- or
tried to, and got the message:

This programme has performed an illegal operation (etc.)

The DETAILS button reveals the following:

PBDLL cause a general protection fault
in module PBDLL.EXE at 0003:0000466f.

[code/]Registers:

EAX=0000037f CS=0dc7 EIP=0000466f EFLGS=00000246
EBX=00004343 SS=30af ESP=0000ad00 EBP=8055ad08
ECX=ffff0000 DS=30af ESI=000077f7 FS=10ff
EDX=00000000 ES=of1f EDI=00000370 GS=of4f

Bytes at CS:EIP:

26 8b 4f 0a 0f b6 d8 b8 1c 72 e8 0b 6c eb 66 80

Stack dump:

00000340 0e400000 3f81ad22 02d03c97 3452362b
006833eb 00050097 0dcf0205 000000bc 00000000[/code]

(plus 6 more of 00000000)

The EIP value is that which is where the gpf supposedly occurred.
Does this mean a bug in PbDLL? What's to do?

------------------