Any idea when the bug in FPC concerning calling functions assigned by doing this:

OnPaint := @FORM1PAINT;
if Assigned(OnPaint) then
   OnPaint(Sender);

This still crashes as of yesterday.....

Thanks
Shane

>> Curtis White 11/24/99 12:08AM >>>

Speaking of keeping backups of earlier compilers....

I have a bash shell script (but it would run under KSH and other shells) that I made that cleans everything up properly and builds the compiler and all the other related parts. It keeps a copy of the compiler. About once a week or so, I just go through and delete some of the older compilers. But by doing it this way, I always have at least the last 5 compilers. So it is easy to use older versions. It is a pretty decent little script for 2 minutes of work. It saves me a ton of time. Anyway, if anyone wants a copy of this script, let me know and I will send it to you.

Thanks...Curtis

_________________________________________________________________ To unsubscribe: mail with
"unsubscribe" as the Subject
archives at http://www.miraclec.com/list_archives/lazarus

Can you let us know when it's fixxed. I changed the code to bypass the problem for now so we may not notice the fix right away.

thanks again
Shane
P.S. Anyone check out genetic-programming.org? Interesting stuff...

>> 11/23/99 04:06PM >>>

Shane Miller wrote

I agree it's in the compiler. I disabled the OPPAINT assignment and the ONDESTROY assignment in lazarus (main.pp) and changed the TControl(data).windowproc to dispatchs and now Lazarus runs! There was something changed in the compiler that has stop functioning. Can someone on the compiler (FPC) end comment so we know you are looking into this please......

There is indeed a bug caused by a patch of Pierre Muller :( We are looking into it, because it disturbs lots of other things as well, please be patient while we try to fix it.

And ALWAYS keep a backup of earlier compilers :-)

Michael.

_________________________________________________________________ To unsubscribe: mail with
"unsubscribe" as the Subject
archives at http://www.miraclec.com/list_archives/lazarus

OK, I haven't really gotten anywhere, but I have more hints.... I am working on hello.pp I did a ppc386 -B -g hello.pp and then I debug it. It was crashing when the signal SHOW was calling the windowproc.

I changed it BACK to DISPATCH and I got further. Now, when I try to close it and do this...

if assigned(FOnDestroy) then
    FOnDestroy(Self);

it crashes on the second line. It never gets into HelloForm OnDestroy function though so it's something with the way it's trying to call that procedure.

Any thoughts? I know this used to work so I'm not sure what changed.... Shane

Shane wrote

can you send the good compiler executable to so I can

try it? I don't have a good copy.

I just built one and seems ok (it's in its way)

Marc

That would be REAL good. I am out of town myself until Sunday so you don't need to rush to get it to me. :-)

==== Programming my first best destiny! ====

Michael A. Hess Miracle Concepts, Inc.
http://www.miraclec.com

I agree it's in the compiler. I disabled the OPPAINT assignment and the ONDESTROY assignment in lazarus (main.pp) and changed the TControl(data).windowproc to dispatchs and now Lazarus runs! There was something changed in the compiler that has stop functioning. Can someone on the compiler (FPC) end comment so we know you are looking into this please......

Shane

>> Marc Weustink 11/23/99 03:24PM >>>

you wrote

OK, something is really messed. I know that the WINDOWPROC call in

gtkcallback used to work because I remember seeing the WRITELN('IN TCONTROL WNDPROC').

Also, I know that the call to ONDESTRY used to work. It seems to be

something that changed in the compiler....is this possible? Does anyone know of anything that would have changed recently that would effect this?

Marc, does the 11/11/99 version of the compiler call these procedures

correctly and does the wndproc in TCONTROL ever get called after compiling with this version?

allright, it took some time (I used the wrong symlink :-) but here are some results of different compilers. It looks like there is something changed in the latest versions (only the 11/11 version is ok). As a final test I'll do a make_really_really_really_really_clean and rebuild all to see if something changes.

Marc

ppc386.1999.11.11:

<snip>
[TCustomForm.CreateWnd] Assign Callback Signals
IN GETFIXED
Exiting GETFIXED
[TCustomForm.CreateWnd] end;
SETTING VISIBLE = TRUE
[TgtkObject.IntSendMessage3] THELLO --> Sent LM_Message: $100E; Data: 0 [TgtkObject.IntSendMessage3] THELLO --> Show/Hide
**************--IN SHOW HIDE
VISIBLE = TRUE
IN SHOW CALLBACK
-----------------IN TCUSTOMFORM WNDPROC------------------- LM_SHOW RECEIVED!!!!!!!!!!!
IN TWINCONTROL WNDPROC
-----------------IN TCONTROL WNDPROC------------------- CALLED THE WNDPROC
**************--Exiting SHOW HIDE
[TWinControl.Show] end;
-----------------IN TCUSTOMFORM WNDPROC------------------- IN TWINCONTROL WNDPROC
-----------------IN TCONTROL WNDPROC------------------- marc:~/fpc/lazarus/examples$

ppc386.1999.11.19:

<snip>
[TCustomForm.CreateWnd] Assign Callback Signals
IN GETFIXED
Exiting GETFIXED
[TCustomForm.CreateWnd] end;
SETTING VISIBLE = TRUE
[TgtkObject.IntSendMessage3] THELLO --> Sent LM_Message: $100E; Data: 0 [TgtkObject.IntSendMessage3] THELLO --> Show/Hide
**************--IN SHOW HIDE
VISIBLE = TRUE
IN SHOW CALLBACK
An unhandled exception occurred at 0x00000000 :
Access violation

Segmentation fault
marc:~/fpc/lazarus/examples$

ppc386.1999.11.22:

<snip>
[TCustomForm.CreateWnd] Assign Callback Signals
IN GETFIXED
Exiting GETFIXED
[TCustomForm.CreateWnd] end;
SETTING VISIBLE = TRUE
[TgtkObject.IntSendMessage3] THELLO --> Sent LM_Message: $100E; Data: 0 [TgtkObject.IntSendMessage3] THELLO --> Show/Hide
**************--IN SHOW HIDE
VISIBLE = TRUE
IN SHOW CALLBACK
An unhandled exception occurred at 0x00000000 :
Access violation

Segmentation fault
marc:~/fpc/lazarus/examples$

_________________________________________________________________ To unsubscribe: mail with
"unsubscribe" as the Subject
archives at http://www.miraclec.com/list_archives/lazarus

That's a good idea. I will do that. I am off work until next Tuesday, so I hope to get some work done on Lazarus, including this. Also, I am still working on putting a web page together with all the changes needed (that I know of) to convert a Delphi program to FPC/Lazarus. I will send it to you when I get it done so you can add it to the site.

Curtis

I haven't found the cause for this yet, but I have been able to change the way it crashes. It seems that if you comment out the Notebook.show in MAIN.PP it acts one way, and with it in it acts another. does this occur with everyone else? One way it crashes with an exception, the other way it hangs.

Any thoughts?

Shane

I think this behaviour is by intention, because the first form is somewhat more readable (and it can't be confused with a call to a function). In Delphi mode ({$MODE Delphi}) the second version should work, so compatiblity isn't really an issue.

cu,
Sebastian

you wrote

OK, something is really messed. I know that the WINDOWPROC call in

gtkcallback used to work because I remember seeing the WRITELN('IN TCONTROL WNDPROC').

Also, I know that the call to ONDESTRY used to work. It seems to be

something that changed in the compiler....is this possible? Does anyone know of anything that would have changed recently that would effect this?

Marc, does the 11/11/99 version of the compiler call these procedures

correctly and does the wndproc in TCONTROL ever get called after compiling with this version?

allright, it took some time (I used the wrong symlink :-) but here are some results of different compilers. It looks like there is something changed in the latest versions (only the 11/11 version is ok). As a final test I'll do a make_really_really_really_really_clean and rebuild all to see if something changes.

Marc

ppc386.1999.11.11:

<snip>
[TCustomForm.CreateWnd] Assign Callback Signals
IN GETFIXED
Exiting GETFIXED
[TCustomForm.CreateWnd] end;
SETTING VISIBLE = TRUE
[TgtkObject.IntSendMessage3] THELLO --> Sent LM_Message: $100E; Data: 0 [TgtkObject.IntSendMessage3] THELLO --> Show/Hide
**************--IN SHOW HIDE
VISIBLE = TRUE
IN SHOW CALLBACK
-----------------IN TCUSTOMFORM WNDPROC------------------- LM_SHOW RECEIVED!!!!!!!!!!!
IN TWINCONTROL WNDPROC
-----------------IN TCONTROL WNDPROC------------------- CALLED THE WNDPROC
**************--Exiting SHOW HIDE
[TWinControl.Show] end;
-----------------IN TCUSTOMFORM WNDPROC------------------- IN TWINCONTROL WNDPROC
-----------------IN TCONTROL WNDPROC------------------- marc:~/fpc/lazarus/examples$

ppc386.1999.11.19:

<snip>
[TCustomForm.CreateWnd] Assign Callback Signals
IN GETFIXED
Exiting GETFIXED
[TCustomForm.CreateWnd] end;
SETTING VISIBLE = TRUE
[TgtkObject.IntSendMessage3] THELLO --> Sent LM_Message: $100E; Data: 0 [TgtkObject.IntSendMessage3] THELLO --> Show/Hide
**************--IN SHOW HIDE
VISIBLE = TRUE
IN SHOW CALLBACK
An unhandled exception occurred at 0x00000000 :
Access violation

Segmentation fault
marc:~/fpc/lazarus/examples$

ppc386.1999.11.22:

<snip>
[TCustomForm.CreateWnd] Assign Callback Signals
IN GETFIXED
Exiting GETFIXED
[TCustomForm.CreateWnd] end;
SETTING VISIBLE = TRUE
[TgtkObject.IntSendMessage3] THELLO --> Sent LM_Message: $100E; Data: 0 [TgtkObject.IntSendMessage3] THELLO --> Show/Hide
**************--IN SHOW HIDE
VISIBLE = TRUE
IN SHOW CALLBACK
An unhandled exception occurred at 0x00000000 :
Access violation

Segmentation fault
marc:~/fpc/lazarus/examples$

OK, my problem is solved. After some websearch I found this error related to color depths and X-servers out of resources.

When I changed the default bitmap depth back to 8bpp (in stead of 24) the editor draws itself :-) But still no text :-(

Marc

Removing the ONDestroy := @?????? fixed the crash and it exits normally now........odd....

Shane

Yep, I did an update today and it appears to work correctly. I will look into it more...

On eother thing. I stilll need to assign TNotifyevent procedures like this:

OnPaint := @FORMPAINT

instead of

OnPaint := FORMPAINT;

Can we get this fixed please. That is a major problem with compatability.

thanks
Shane

>> Sebastian G�nther 11/30/99 08:54AM >>>

??? Are you sure? KCL worked after I updated the compiler (after Florians announcement that this should work again)

- Sebastian

_________________________________________________________________ To unsubscribe: mail with
"unsubscribe" as the Subject
archives at http://www.miraclec.com/list_archives/lazarus

Speaking of keeping backups of earlier compilers....

I have a bash shell script (but it would run under KSH and other shells) that I made that cleans everything up properly and builds the compiler and all the other related parts. It keeps a copy of the compiler. About once a week or so, I just go through and delete some of the older compilers. But by doing it this way, I always have at least the last 5 compilers. So it is easy to use older versions. It is a pretty decent little script for 2 minutes of work. It saves me a ton of time. Anyway, if anyone wants a copy of this script, let me know and I will send it to you.

Thanks...Curtis

Shane,

I tried changing things yesterday. I was trying to debug hello_world.pp. I crashes on me when windowproc is called from gtkshowCB in gtkcallback.inc. I changed the call to windowproc back to Dispatch and the hello window displayed but no close button. It still cored when I exited. I briefly studied how windowproc is used in Delphi and the is seems the lcl is lacking a full implementation of windowproc...(vs directly calling Dispatch)

Tom

OK, something is really messed. I know that the WINDOWPROC call in gtkcallback used to work because I remember seeing the WRITELN('IN TCONTROL WNDPROC').

Also, I know that the call to ONDESTRY used to work. It seems to be something that changed in the compiler....is this possible? Does anyone know of anything that would have changed recently that would effect this?

Marc, does the 11/11/99 version of the compiler call these procedures correctly and does the wndproc in TCONTROL ever get called after compiling with this version?

Shane

>> "Shane Miller" 11/23/99 01:56PM >>>

Removing the ONDestroy := @?????? fixed the crash and it exits normally now........odd....

Shane

_________________________________________________________________ To unsubscribe: mail with
"unsubscribe" as the Subject
archives at http://www.miraclec.com/list_archives/lazarus

??? Are you sure? KCL worked after I updated the compiler (after Florians announcement that this should work again)

- Sebastian

+ From: Shane Miller
+
+ I haven't found the cause for this yet, but I have been able
+ to change the way it crashes. It seems that if you comment
+ out the Notebook.show in MAIN.PP it acts one way, and with it
+ in it acts another.
+ does this occur with everyone else? One way it crashes with
+ an exception, the other way it hangs.
+
+ Any thoughts?

After I got the latest version of all (to build a new compiler), my original crash in Font.Assign is gone, however now after I try to place text in the editor (by pressing the topmopst button in Testform) I get something like:

Gdk-ERROR **: BadValue (integer parameter out of range for operation) serial 1556 error_code 2 request_code 53 minor_code 0

In the debugger you can see that the button-press event is completely handled, the callback returns and thist is the result. When the text is placed, the editor in invalidated through a call to gtk_widget_queue_draw. I (wild ?) guess that this draw is executed as soon as the gtkclickedCB (?) is finished.

I think the key to this problem is the message I get... but what does it mean. Anyone any clue ?

Marc

Shane Miller wrote

I agree it's in the compiler. I disabled the OPPAINT assignment and the ONDESTROY assignment in lazarus (main.pp) and changed the TControl(data).windowproc to dispatchs and now Lazarus runs! There was something changed in the compiler that has stop functioning. Can someone on the compiler (FPC) end comment so we know you are looking into this please......

There is indeed a bug caused by a patch of Pierre Muller :( We are looking into it, because it disturbs lots of other things as well, please be patient while we try to fix it.

And ALWAYS keep a backup of earlier compilers :-)

Michael.

you wrote

I haven't found the cause for this yet, but I have been able to change the

way it crashes. It seems that if you comment out the Notebook.show in MAIN.PP it acts one way, and with it in it acts another.

does this occur with everyone else?

I haven't tried the latest compiler (today) but with the one of the 18/19th I've troubles, the one I built on 11 gives no troubles at all.

can you send the good compiler executable to so I can try it? I don't have a good copy.

Shane

>> Marc Weustink 11/22/99 12:22PM >>>

you wrote

I haven't found the cause for this yet, but I have been able to change the

way it crashes. It seems that if you comment out the Notebook.show in MAIN.PP it acts one way, and with it in it acts another.

does this occur with everyone else?

I haven't tried the latest compiler (today) but with the one of the 18/19th I've troubles, the one I built on 11 gives no troubles at all.

_________________________________________________________________ To unsubscribe: mail with
"unsubscribe" as the Subject
archives at http://www.miraclec.com/list_archives/lazarus

How about checking your script against the one supplied in the HowTo for Linux and seeing if you can either tweek that one or just exchange it with yours. Send me the modifications to the HowTo and I'll just add this to the web page.