I'm migrating a big Application from PowerBuilder 5.04 to 6.5.1.
There was no problem in migrating the *.pbl.

When Creating the native Executable for Windows NT 4.0, there is a Dr.
Watson in pb60.exe (Access Violation).

The compiler switches are set the following way:
1) Rebuild -> Full
2) Machine Code -> Yes (Checked)
3) Trace Information -> No (Not Checked)
4) Error Context Information -> No (Not Checked)
5) Executable Format -> Native
6) Optimization -> Speed
7) For each Library has to be created a dll. Therefore the checkbox is
checked for each Library.


I localized one Problem in PB6.5 and PB6.5.1 in a simple testprogram (only
two lines of code):
Do the following to reproduce the problem:

Step-1: Create a new, empty powerBuilder Project.
Step-2: Create a new Window.
Step-3: Code just the following tow lines in the Open Event of the Window:

        String    ls_test
        ls_test = "#�#"

Step-4: Try to create the native executable with the switches mentioned
above and you will have an access violation in PB60.exe.


Questions:
1) What's wrong with PB6.5 and PB6.5.1?
2) Is there a work around? (I really need native Executables and a dll for
each pbl, so creating *.pbd cannot be the answer!)
3) Is this problem still existent in PB7.0?

Thanks for help

Andy
aj@active-software.ch

Running Win98, PB 6.5.1, no abnormal aborts at all.

Andreas Jost <aj@active-software.ch> wrote in message
news:Nojrrhtv#GA.301@forums.sybase.com...
I'm migrating a big Application from PowerBuilder 5.04 to 6.5.1.
There was no problem in migrating the *.pbl.

When Creating the native Executable for Windows NT 4.0, there is a Dr.
Watson in pb60.exe (Access Violation).

The compiler switches are set the following way:
1) Rebuild -> Full
2) Machine Code -> Yes (Checked)
3) Trace Information -> No (Not Checked)
4) Error Context Information -> No (Not Checked)
5) Executable Format -> Native
6) Optimization -> Speed
7) For each Library has to be created a dll. Therefore the checkbox is
checked for each Library.


I localized one Problem in PB6.5 and PB6.5.1 in a simple testprogram (only
two lines of code):
Do the following to reproduce the problem:

Step-1: Create a new, empty powerBuilder Project.
Step-2: Create a new Window.
Step-3: Code just the following tow lines in the Open Event of the Window:

        String    ls_test
        ls_test = "#�#"

Step-4: Try to create the native executable with the switches mentioned
above and you will have an access violation in PB60.exe.


Questions:
1) What's wrong with PB6.5 and PB6.5.1?
2) Is there a work around? (I really need native Executables and a dll for
each pbl, so creating *.pbd cannot be the answer!)
3) Is this problem still existent in PB7.0?

Thanks for help

Andy
aj@active-software.ch

There is an open CR on this (166106).  The problem appears to be with
international characters, only when compiling to machine code and,
for whatever reason, only when compiling on NT.  If you compile it on
a Win95 machine it will work fine.

On Fri, 25 Jun 1999 09:07:24 +0100,
 in powersoft.public.powerbuilder.general
Andreas Jost <aj@active-software.ch> wrote: 
>I'm migrating a big Application from PowerBuilder 5.04 to 6.5.1.
>There was no problem in migrating the *.pbl.
>
>When Creating the native Executable for Windows NT 4.0, there is a Dr.
>Watson in pb60.exe (Access Violation).
>
>The compiler switches are set the following way:
>1) Rebuild -> Full
>2) Machine Code -> Yes (Checked)
>3) Trace Information -> No (Not Checked)
>4) Error Context Information -> No (Not Checked)
>5) Executable Format -> Native
>6) Optimization -> Speed
>7) For each Library has to be created a dll. Therefore the checkbox is
>checked for each Library.
>
>
>I localized one Problem in PB6.5 and PB6.5.1 in a simple testprogram (only
>two lines of code):
>Do the following to reproduce the problem:
>
>Step-1: Create a new, empty powerBuilder Project.
>Step-2: Create a new Window.
>Step-3: Code just the following tow lines in the Open Event of the Window:
>
>        String    ls_test
>        ls_test = "#�#"
>
>Step-4: Try to create the native executable with the switches mentioned
>above and you will have an access violation in PB60.exe.
>
>
>Questions:
>1) What's wrong with PB6.5 and PB6.5.1?
>2) Is there a work around? (I really need native Executables and a dll for
>each pbl, so creating *.pbd cannot be the answer!)
>3) Is this problem still existent in PB7.0?
>
>Thanks for help
>
>Andy
>aj@active-software.ch
>
>
>
>

---
Bruce Armstrong [TeamSybase]            | Romac/Source International
mailto:Bruce.Armstrong@teamsybase.com   | mailto:jobs@sourcela.com
                                        | http://www.romac-source.com

Preach the gospel at all times. If necessary, use words. [Francis of Assisi]
http://www.kidbrothers.org                http://www.fccwc.org
http://www.harvest.org/knowgod/index.htm

-----------== Posted via the PFCGuide Web Newsreader ==----------
http://www.pfcguide.com/_newsgroups/group_list.asp

Are you doung a lot calculations, etc.?
What is online refreshing? GUI refreshing? If yes then dll's will not do you
any good...
Andreas Jost <aj@active-software.ch> wrote in message
news:1q#jG1xv#GA.206@forums.sybase.com...
> Hi Philip
>
> The reason, why we build native dll is speed!
> Our Application runs faster using native executables and we really need
this
> speed for online refreshing etc.
>
> Philip Salgannik wrote in message ...
> >
> >Andreas Jost <aj@active-software.ch> wrote in message
> >news:Nojrrhtv#GA.301@forums.sybase.com...
> >> I'm migrating a big Application from PowerBuilder 5.04 to 6.5.1.
> >> There was no problem in migrating the *.pbl.
> >>
> >> When Creating the native Executable for Windows NT 4.0, there is a Dr.
> >> Watson in pb60.exe (Access Violation).
> >>
> >> The compiler switches are set the following way:
> >> 1) Rebuild -> Full
> >> 2) Machine Code -> Yes (Checked)
> >
> >Is there a specific reason for #2 ?
> >
>
>

Hi Philip

The reason, why we build native dll is speed!
Our Application runs faster using native executables and we really need this
speed for online refreshing etc.

Philip Salgannik wrote in message ...
>
>Andreas Jost <aj@active-software.ch> wrote in message
>news:Nojrrhtv#GA.301@forums.sybase.com...
>> I'm migrating a big Application from PowerBuilder 5.04 to 6.5.1.
>> There was no problem in migrating the *.pbl.
>>
>> When Creating the native Executable for Windows NT 4.0, there is a Dr.
>> Watson in pb60.exe (Access Violation).
>>
>> The compiler switches are set the following way:
>> 1) Rebuild -> Full
>> 2) Machine Code -> Yes (Checked)
>
>Is there a specific reason for #2 ?
>

Yes, we do CUI-Refreshing quite intensive (including calculations).
Our Object Hirarchie has many levels of inheritance. We make use of our own
framework.
Therefore we do not use pfc in this project.

At all, why do you think dlls will not do any good?

Thanks for the hint. So I will install a Win95 machine to try again.

I played round with this problem on a NT Workstation last night.
I managed to create the exe and the dll for the described testprogram (on NT
4.0).
Maybe it helps to solve the real problem at Sybase. So far, I did not check,
if it's possible to create the exe and dlls for our real application which
consists not only 2 lines of code :-)
I tried the following (on NT):

1) Create the exe in p-code. Create a pbd for each pbl.
2) Create another exe instantly after step-1. Create Machine code this time!
3) => The exe and the dll was created!!!

It could be, that there is a library missmatch in the Installation-Program
of PowerBuilder or
that there is a problem with a selfregistrating dll in the setup?



Bruce Armstrong [TeamSybase] wrote in message ...
>There is an open CR on this (166106).  The problem appears to be with
>international characters, only when compiling to machine code and,
>for whatever reason, only when compiling on NT.  If you compile it on
>a Win95 machine it will work fine.
>

So it seems to be a real NT Problem.
By the way, is PowerBuilder supported for Win98? (I thought not officially).


Lu Powell wrote in message <6n45ecxv#GA.287@forums.sybase.com>...
>Running Win98, PB 6.5.1, no abnormal aborts at all.

On Sat, 26 Jun 1999 14:13:36 +0100,
 in powersoft.public.powerbuilder.general
Andreas Jost <aj@active-software.ch> wrote: 
>Maybe it helps to solve the real problem at Sybase. So far, I did not check,
>if it's possible to create the exe and dlls for our real application which
>consists not only 2 lines of code :-)

Well, bear in mind I'm not a Sybase employee, I'm an end-user just like you.

>I tried the following (on NT):
>
>1) Create the exe in p-code. Create a pbd for each pbl.
>2) Create another exe instantly after step-1. Create Machine code this time!
>3) => The exe and the dll was created!!!

Wild.  Perhaps that info would help them track down the bug.  However, they
have been able to reproduce the original problem, so I'm sure that they're
working on it.  I just don't know what priority it has.


---
Bruce Armstrong [TeamSybase]            | Romac/Source International
mailto:Bruce.Armstrong@teamsybase.com   | mailto:jobs@sourcela.com
                                        | http://www.romac-source.com

Preach the gospel at all times. If necessary, use words. [Francis of Assisi]
http://www.kidbrothers.org                http://www.fccwc.org
http://www.harvest.org/knowgod/index.htm

-----------== Posted via the PFCGuide Web Newsreader ==----------
http://www.pfcguide.com/_newsgroups/group_list.asp