"EXE has stopped working ..."

[arnoldhf]

I have started converting 3 of a family of inter-related QB programs to QB64 and have run into the issue of getting the Windows message:

"program" has stopped working

A problem caused the program to stop working correctly.
Windows will close the program  etc, etc.

The message is not immediate but comes up within a minute on all 3. 

Each program can start another using RUN. Sometimes the message appears when another program loads but before the program itself displays its screen.

Within one program I do some data file access and after pulling up records a few times the error appears.

Any suggestions on how to diagnose or eliminate the issue?

Thanks,

[bplus]

Instead of RUN use Shell and Shell a compiled bas. You can Compile a bas and Shell that exe (to Run it) on the fly if you are careful with folders.

[arnoldhf]

Thanks. Tried it:

Replaced RUN with SHELL. It didn't solve problem (still got WIN error) and when SHELLING the calling program remains open which I don't want since going between programs leaves multiple iterations of programs still open.

Also, I am not sure what you mean by "Shell a compiled bas." I am compiling using QB64.exe and running the EXEs.

(FYI the original QB programs used CHAIN because of common variables but I removed CHAIN and used RUN, but that's going to be a question for a different thread.)

A

[SMcNeill]

Can you share the code so we can test the issue ourselves?

[mdijkens]

I'd say start by integrating all 3 sources in one.
Just change overlapping labels/linenumbers or break up in easier to maintain subs and functions.

Chain was used because of memory limitations.
Now with QB64 this is virtually gone.

[arnoldhf]

Thank you for the responses. I did "solve" the problem but I want to reply to you both first.

To SMcNeill:
                    It would not be beneficial to give the source code since the programs require a specific directory structure and data files to execute properly. Also, what turned out to be the offending program is probably around 4000 lines long.

To mdijkens:

A bit of background on the programs would be useful here:

I started developing a "suite" of interrelated programs starting in the mid 80's. Because of QB program size restrictions I used $INCLUDE often as well as compiling multiple .BAS modules at into a single EXE.

The result has been about 60+ fairly large BAS programs compiled into around 40+ EXEs all using CHAIN (to go to other modules) for COMMON variables. I will probably be changing that.

To see if converting to QB64 was a job I wanted to tackle I "converted" 3 programs enough to run. So to your point, trying to combine them (all 60+) into one module was a thought I had but there are so many overlapping labels and line numbers as well as contradictory/reused variable names that I would give up if that was the only route.

--->
Now as to the original reason for this thread, after much insertion of my own debugging code and many recompilings of the larger of the 3 programs I was able to narrow the cause of the error down to several lines of code:

FIELD #10, 80 AS LG$
also
GET #1, rn
and perhaps I will still find others.

Strangely I didn't get the error every time one of these lines was executed.

If I REM these lines the error disappears completely. So it had something to do the the file I/O or access.
I rewrote some code before these lines which OPENed the file, unREMed the offending lines, and now it works fine.

I don't know why I didn't just get a more specific RUN TIME I/O error with error line # instead of the generic and extremely unhelpful Windows "program stopped responding" message.

Again, thanks for the replies. Will probably post more questions later.

A

[mdijkens]

Thanks for your clarification.
Been there also in the 80's; used same techniques.
Your errors are strange and non-saying indeed
With refactoring of much of my old ('80-'90) code I've, in a lot of cases, rewritten the file access parts too to get things reliable.
Another point of attention is serial