Call for BUGS.doc 06-May-00


From: "Willie Moore" 
Subject: Call for BUGS

I have seen a lot of threads on repo stability, etc on the
forums. CA World is comming up next week.  The entire
development team and a lot of the technical support team
will be in new Orleans.

If you have any reproducable steps/samples that you can
email me (williem@bigfoot.com), I will be happy to get in
their ear. The developers are always willing to look,
listen, and FIX if you can present a reproducible error.

Marijo was polling for a reproducible repo error on the CA
Forum yesterday.

Regards,

Willie


From: DEarl@NewYorker.DE (David Earl)
Subject: Re: Call for BUGS

OK, one of my mickey mouse specials.  Set focus on the error
browser, through the Tools menu, by actually generating an
error during compile, whatever.  (You may now fix the error
if you like.)  From the error browser, press +.
It doesn't work from that one window.  Anywhere else in the
IDE, that hotkey will run the compiler.

David.

From: "Geoff Schaller" 
Subject: Re: Call for BUGS

Hi Ginny,

I did that too - also reproduced again in my recent post to
the Open Forum and here. The trouble is that it doesn't
always happen on cue . Typical of the worst kind of bugs.
But at least the procedure is as I and 8 others know it has
been recorded.

The reason for sending the repo is that because it fails to
delete or recompile, it is easier to pick up on the error
condition that has occurred. It is in the indexes and
according to Paul Piko, should be trackable. This is a
little like reverse engineering. Find the bits that went
wrong and look for the code that affects them. As it is
totally repo-internal this sort of reverse debugging should
work. Or at least I hope so .

Geoff


From: "Pavel Vetesnik" 
Subject: Re: Call for BUGS

I have 2 things, which are not really 'bugs' but my
suggestion:

1. DateTimePicker:Value is replaced with NIL when the owner
window closes. So there is no way how to use DateTimePicker
in the modal dialogs which should return the value back. At
least I don't know any.

2. Close() and QuerryClose() methods are not called for
windows owned by tab pages (when the main window closes).

3. In the help (or somewhere else) put message what version
of ComCtl is required.

--
Pavel


From: "Geoff Schaller" 
Subject: Re: Call for BUGS

Pavel,

I have seen workarounds in the NG for your point (1) but my
understanding is that this is a limitation of the control
itself. If you can't find one, I can send you those I have
saved in my VO Wisdom.

Point (3) is a valid and interesting one.

But point (2), I think, deserves further comment. The same
"issue" can be extended to subdatawindows. I think, and I
stress it is IMO only, that this is because the closure of
such windows (which are usually derivatives of datawindows)
should be the responsibility of their owner. The last thing
you want is for that child window closing itself without
care for servers and/or the resulting functionality of the
owner window. Usually, you close the owner and also close
all the children but the children are not closed independent
of the owner. Hence Close() and QueryClose() are not
required. But you still have Destroy() so you can still
remove resources and close additional servers if you feel
strongly about such things.

What do you think?

Geoff


From: "Mathias Hakansson" 
Subject: SV: Call for BUGS

This generates strange errors if you have an ocx control in
that window. In that case the parent window must call
Endwindow in it's children to avoid errors. I also think you
have the same problem in splitwindows.


Mathias


From: "Geoff Schaller" 
Subject: Re: Call for BUGS

OK, but for subdatawindows you will need to make that call
explicitly.

Geoff


From: "Will Chapman" 
Subject: Re: Call for BUGS

Could someone confirm this one:

1. Open up a piece of code in the Source code Editor.
2. Highlight a phrase and open up the search feature by
clicking on the magnifying glass with an A in it (no red)
and then open up 'Advanced'.
3. Select Current Application and then Find Next.
4. This will give you a selection of code lines, double-
click on one and edit a lin in the code (it can be a dummy
edit).
5. Close the instance of the Source code editor  that
displays the edited code.
6. On my system I VO2.5a1 crashes with an error in
cavoedit.dll.

Cheers..

Will Chapman

From: ernestomaff@my-deja.com
Subject: Re: Call for BUGS

Hi Will,

I just tested following your instructions and I had the same
result as you experience. Same error message!

Regards
Ernesto


From geoffsch@bigpond.net.au Sat Apr 01 21:39:50 2000
From: "Geoff Schaller" 
Subject: Re: Call for BUGS

Hi Will,

I concur also.

There are 4-5 variations of this bug, unfortunately. The
workaround for this is to use VOPP .

Geoff


From: "Pavel Vetesnik" 
Subject: 5 steps to corrupt repository

I've found a way (one of the ways, I guess) how to get
repository corrupted. It works 100% on my computer, I'd like
to ask somebody to follow those steps and report results:

1. Create menu and at least one menu item has an
accelerator.
2. Save the menu and open the generated accelerator resource
in source code editor.
3. Manually edit something inside (change one key, for
example).
4. Save accelerator resource.
5. Open again the menu (in menu editor) and change something
inside (the accelerator key, for example) and try to save
it.
-> Unhandled exception error appears, VO closes.
-> after VO is opened again, it is not possible to build the
application (xxx dead entities).
-> external reindex helps (a bit), it is possible to build
application now.
-> but when you try to open the menu (in menu editor) and
make some changes it crashes again.
-> you may externally reindex it how many times you want,
menu editor for that menu always crashes.

I hope I was clear. I tried it several times (even on new
project and with empty application) and got always the same
results. It was on VO2.5a-1, Windows2000.

TIA
--
Greetings from
Pavel Vetesnik,
Prague (Czech republic)


From: "Geoff Schaller" 
Subject: Re: 5 steps to corrupt repository

Yes Pavel, I agree!

In fact, I have noticed that often, two external reindexes,
one immediately after the other, are required to fix VO-
created repo corruptions. If you re-enter VO without doing
the second reindex, it seems to ignore you and goes back to
the corruption. I wonder what is happening for two reindexes
in a row to be of importance?

Geoff


From: Philippe Legros 
Subject: Re: 5 steps to corrupt repository

Pavel,

I reproduced the problem, but not entirely. After I
restarted VO, I was always able to build an run the app,

without having to re-index. I run Win98.

Another precision: the problem occurs when I save the MED by
closing it. If I save first with the save menu, the close
it, no error.

PhL


From: 100725.3150@compuserve.com (Karl Faller)
Subject: Re: Call for BUGS

Willie,

possibly not new, also no sample, sorry, only a how to :

New entity =>DBServerEditor ( Default to CDX )
type Classname, Filename, Indexname; Enter, =>Indexname gets
transfered to Ordername. Click in Orders. Enter. = Newline,
type second Ordername, Enter => Propertywindows is blank
(ie. no Ordername ) Click PropWin, type secondOrdername by
hand, save, exit Editor, reopen => no second Order exists!!

The interesting thing is, IF you add at least _one_
fieldname _ BEFORE typing the indexname, all is ok. So it's
not _so_ bad, but IMHO shows that in the Editor not
everything is quite "sound"....

HAND
Karl


From: David_wright@spra.com (David Wright)
Subject: Some Bugs? ( was Re: Call for BUGS)

Here are a list of some minor bugs and a few annoyances.
I am using CA-VO 2.5, Windows 95. Don't ask why.

A. Source Code Editor:
1. Well, its probably as good as the Visual Basic editor!
2. File=>Exit in the Editor exits the entire program, not
the editor.
3. My Microsoft Intellipoint Wheel (third button or wheel)
mouse makes CA-VO crash when I use the wheel in the popup
help for objects (object:yada yada).
4. The filename of the file you are editing is not in the
title bar of the window.
5. A mouse click in a selection will not turn off the
selection.

B. DbServer Editor:
1. It appears unable to handle certain numeric indexes
(converted to strings). Same with CreateIndex() (DBF class).
Specifically, I tried creating an index with the indexing
expression Str(x,1,0)+Str(y,1,0) and various combinations
thereof.

I was able to create the index in a third party dbf editor,
cmVodbx32, using the same code(copy and pasted the code).

2. If you don't have an index in your database, and you
generate the code, the program will define a local variable
llindexfile, which is never used, generating a compiler
error if you have the appropriate compiler option selected
(type inferences).

C. Help:
Sleep() is not there.
ApplicationExec() is not there.

D. IDE
Selecting to hide "Fields" from view also hides "Forms".

E. Selecting "Sorted" in a ComboListBox that is filled with
a Server database will ignore items that are out of sort
sequence. The listbox will display records "A","B","C","D"
out of a records with strings "A","B","C","D","BX".

From: "Geoff Schaller" 
Subject: Re: Some Bugs? ( was Re: Call for BUGS)

Hi David,

(..and yes, you are crazy sticking to W95 - sorry, couldn't
resist )

> A. Source Code Editor:
> 2. File=>Exit in the Editor exits the entire program, not
the editor.
This is deliberate and I find it quite handy.

> 3. My Microsoft Intellipoint Wheel (third button or wheel)
mouse makes CA-VO crash when I use the wheel in the popup
help for objects (object:yada yada).
Have not noticed that - try reinstalling your mouse driver.

> 4. The filename of the file you are editing is not in the
title bar of the window.
Do you mean the "entity" name, because the module name
certainly is.
However, maybe it would be a useful "enhancement" but
couldn't be classified as a bug.

> 5. A mouse click in a selection will not turn off the
selection.
??? Somehow I think this is possibly a personal preference.
I don't want it to or I would be continually be re-making
the selection from jittery hand movements.

> B. DbServer Editor:
> 1. It appears unable to handle certain numeric indexes
(converted to strings). Same with CreateIndex() (DBF class).
Specifically, I tried creating an index with the indexing
expression Str(x,1,0)+Str(y,1,0) and various combinations
thereof.
I suspect you are really experiencing type casting issues
(and/or setdecimal issues). This expression (which I do not
have trouble with) will convert differently depending on a
whole raft of considerations. Correct me if I'm wrong but
isn't cmVodbx32 strongly typed? The more precise you are

with your "typing", the more consistent the outcome.

> 2. If you don't have an index in your database, and you
generate the code, the program will define a local variable
llindexfile, which is never used, generating a compiler
error if you have the appropriate compiler option selected
(type inferences).
Sure. Its an acknowledged issue and Marijo said twice that
they would change this . However, it arises from a
routine that tries to be "generic".
Simply remove the line from the generated code.

> C. Help:
> Sleep() is not there.
Yes it is - Put the cursor on Sleep and Shift-F1 it. Its an
api.

> ApplicationExec() is not there.
Look up Exec() instead. this is a function wrapper to this
call.

> D. IDE
> Selecting to hide "Fields" from view also hides "Forms".
Don't know what you mean here. I can't find that option in
the View options list. Which "level" are you looking at?

> E. Selecting "Sorted" in a ComboListBox that is filled
with a Server database will ignore items that are out of
sort sequence. The listbox will display records
"A","B","C","D" out of a records with strings
"A","B","C","D","BX".
I don't even try to use the "server" option of a combo box
because I have only ever met with disaster and slow
performance. But I also can't imagine how the combo box is
going to "sort" a dbf? What if the DBF has an index? I would
need to consult the SDK on that one.

Geoff


From: "Geoff Schaller" 
Subject: Re: Some Bugs? ( was Re: Call for BUGS)

Wong,

No way!  An upgrade to W98 is only US70 or thereabouts. With
it comes free internet sharing, modem sharing, sharper
graphics, better internet integration and whole host of
other "features". Plus, the Windows Resource Kit that you
had to pay for in W95 (user and network management, profile
management etc) is all free and standard on W98. It also
fixed many bugs in network management.

Now for the same machine and the same RAM, W98 is no slower
to run but takes about 30 sec longer to boot. Put 64MB RAM
in your PC and it will be fine. Certainly heaps better than
W95 (or are you going to tell me that you cannot afford $30
for 32MB more RAM ?? ).
Sorry, I don't mean any offence either but I don't find your

point of view believable. I am running W98 on a 686 at
166Mhz and its fine. I did use it to develop on once, but
that was some time ago. Productivity is important to me so I
have graduated to a faster mchine.

Cheers.
Geoff

From: "Lars-Eric Gisslen" 
Subject: SV: Some Bugs? ( was Re: Call for BUGS)

Wong,

According to my experiences it's not always as easy as to
just plug in one more SIM/DIM. I have faced problems with
the memory if you mix SIMs from different manufacturers.
Sometimes it's solved if you just change places of SIMs and
sometimes you have to replace the SIMs so all is from the
same menufacturer. I don't know if this depends on the
chipset on the motherboard or not, but it can be problems to
just add one more SIM.

--
Regards,
Lars-Eric
http://www.clippersweden.se

From: nswong@maxis.net.my (Wong)
Subject: Re: SV: Some Bugs? ( was Re: Call for BUGS)

Hi Lars-Eric,

From year 1988 to 1991, I work as computer repairing
technician, every day we repair quite a lot of PC, at that
period we learned a lot of hardware related thing.

If I'm not wrong, this is nothing to do with methorboard
chipset, but are because the two RAM chip does not insync, I
still remember by that time, even RAM chip from same
menufacturer, same model, but different batch, may also
facing this problem.

Anyway, thank for your reminder. :-)

Regards,
Wong

From: David_wright@spra.com (David Wright)
Subject: Re: Some Bugs? ( was Re: Call for BUGS)

>> D. IDE
>> Selecting to hide "Fields" from view also hides "Forms".
>Don't know what you mean here. I can't find that option in
the View options list. Which "level" are you looking at?

that is options, view by enity. My "fields" and "forms" are
reversed.

>> E. Selecting "Sorted" in a ComboListBox that is filled
with a Server database will ignore items that are out of

sort sequence. The listbox will display records
"A","B","C","D" out of a records with strings
"A","B","C","D","BX".
>I don't even try to use the "server" option of a combo box
because I have only ever met with disaster and slow
performance. But I also can't imagine how the combo box is
going to "sort" a dbf? What if the DBF has an index? I would
need to consult the SDK on that one.

Well, I had forgotten that I selected this 'option' in my
early weeks of programming. I spent hours until I figured
this one out.  Perhaps it has a sublte purpose somehow.

From: "Geoff Schaller" 
Subject: Re: Some Bugs? ( was Re: Call for BUGS)

Hi David,

> that is options, view by enity. My "fields" and "forms"
are reversed.
OK - I see what you mean now. And fields does nothing (that
I could see, anyway). I guess I don't use these features so
didn't notice.

> Well, I had forgotten that I selected this 'option' in my
early weeks of programming. I spent hours until I figured
this one out.  Perhaps it has a sublte purpose somehow.
Don't know - still waiting on explanations from CA. But I'm
not holding my breath .

Geoff


From: DEarl@NewYorker.DE (David Earl)
Subject: Re: Some Bugs? ( was Re: Call for BUGS)

>5. A mouse click in a selection will not turn off the
selection.

Actually that seems to be intentional, because you can then
drag that selection somewhere.  I've actually found this one
to be _extremely_ annoying, because of accidentally dragging
code somewhere.

David.

From: michaelyacobucci@my-deja.com
Subject: Re: Call for BUGS

I have worked with a very large VO app for years and I
understand how much time can be wasted on a corrupt
repository. Lately I have not had nearly as many problems
and there is only one explanation. Windows 2000. I have been
running win 2000 since the beta 3. I was able to re-create a
hang in VO. By getting out and right back in of the
repository. Windows 98 hard hung and I had to pull the plug.
I was used to rebooting on a regular basis with both win95
and win98. After the hang I loaded win 2000 and have had
about maybe 10% of the corrupt repository problems that I
had before. I never have to reboot anymore.

Also, I think the NTFS file system makes win 2000 more
stable. You can move your data over without a reformat on
your drive. As far as I know now the only time I have seen a
corrupt repository is if you delete an app in the project.

Hope it helps,

Mike Yacobucci


From: Jean-Marie Berthiaume <#jiembe@#videotron.ca>
Subject: Re: Call for BUGS

I'm using Win95 and the hanging problem disappeared with the
2.5a patch. In fact it only hang once since instead as more
than twice a day at least before.

Jean-Marie Berthiaume

    Source: geocities.com/n_s_wong/vo/ng

               ( geocities.com/n_s_wong/vo)                   ( geocities.com/n_s_wong)