Weird bug...

Hi all,

I am experiencing a very weird bug. Both my computers run Win10 and I develop programs that use ClearWin+ with version 8.10 of FTN95. On just one of them, the same program cannot run - don't even open up the graphics window; on the other, it runs as expected.

If I run the debugger on the offending computer, I get a message 'invalid floating point operation' with address Unknown in wslbscr32.dll

The F95 program just opens up a window with the following commands

rslt=winio@('%ww%ca[Geom -Renderiza]%gr[black,rgb_colours]',tela%xmax-tela%xmin+1,tela%ymax-tela%ymin+1)
rslt=winio@('%lw',ctrl)

Any clues? This is really pestering me right now...

Cheers Rudnei

You appear to have left off the & on first call. Also I would question the type of width and height. I would recommend integer*4 rslt, width, height, ctrl integer(7) handle width = tela%xmax-tela%xmin+1 height = tela%ymax-tela%ymin+1 rslt = winio@('%ww&') rslt = winio@('%ca[Geom -Renderiza]&') rslt = winio@('%gr[black,rgb_colours]&', width, height) rslt = winio@('%lw&',ctrl)
rslt = winio@('%hw',handle)

Hi,

Sorry, my typo above: here is the code pasted from the program:

rslt=winio@('%ww%ca[Geom - Renderiza]%gr[black,rgb_colours]&',tela%xmax-tela%xmin+1,tela%ymax-tela%ymin+1)
rslt=winio@('%lw',ctrl)

and all the variables are of the appropriate type (INTEGER)...

I would still simplify the arguments as suggested and see what happens.

Hi John,

Did that and the problem still happens, it runs once (without any problem) and then subsequent runs do not work - it just hangs, CPU usage goes up to 32% and then decreases to 0% and it is terminated by the OS.

But I did some more testing and found that the same problem is happening with other programs compiled with FTN95 (either Debug w32 or Release w32) that do not use ClearWin+.

I also tried installing the modified .DLLs that were mentioned on a post on this forum and did not fix the problem.

So the problem has to be with Windows 10 or some of its libraries... do you happen to know a related problem and how to fix it?

My computer (Intel i5-6400T with 6GB RAM) is running Windows 10 Home Single Language (Portuguese), 64 bits, version 1703, Build SO 15063.332

Hopefully...

Further to the above: the other computer where the FTN95 programs are running fine has an Intel i5-7200U with 8GB RAM and runs Windows 10 Home, 64-bit OS, version 1607, OS Build 14393.1198.

The versions and builds are different, but that shouldn't be happening...

Rudnei

I was going to suggest that you check if you set ctrl = 0, however you are reporting 'the same problem is happening with other programs compiled with FTN95', possibly non clearwin+ programs ?

I have not seen this happening on my programs or reported on the Forum.

Could you be calling %lw in a subroutine, where 'ctrl' is a local variable? If so, you should move this variable 'ctrl' to COMMON or a MODULE or give it the SAVE attribute. Eddie has described this problem in the past. I am not sure why this would be the case, as if you are using %lw, then you need access to ctrl to use it to close the window later, so it should not have been a local variable.

Two things...

1. Check that rslt and winio@ are integer and function returning integer.

2. See if you can find out what wslbscr32.dll relates to. I can't find it via Google. Does it have any associated drivers and are they up-to-date?

Hi,

I am glad to report that there is nothing wrong with my programs or FTN95 for that matter. 😉

I've traced the problem to an as yet not understood interaction between my programs and the Warsaw internet banking security program which my bank uses.

I've found it out because the other computer on which the programs were running installed a new version of that Warsaw program and immediately my programs failed to run exhibiting the same behaviour as on the other computer. I've uninstalled the Warsaw program and lo and behold my programs were right on track again!

Thank you all for your suggestions!

Cheers Rudnei

Not at all surprised by this. Avast antivirus stops FTN95 programs from running. I'm sure they aren't alone. Kaspersky seems best, and McAfee stops connection to my networked Epson multifunction inkjet printer, but not to my networked Lexmark laser printer and doesn't affect FTN95-compiled programs.

For preference, but for obvious reasons not all the time, I run analyses on a machine not connected to the internet, and not running anything but the barest Windows.

Eddie