🚀 go-pugleaf

In article <4diqjd$b19@macondo.dmu.ac.uk>, c2bs@dmu.ac.uk (Bimal Shah) writes:
|> Hello,
|> 
|> The following is happening in one of the modules
|> i have wrote :
|> 
|> IF condition THEN
|> 	WriteString('blah, blah, blah'); 
|> 	CALLING PROCEDURE HERE
|> 	WriteString('blah, blah, blah');
|> ELSE
|> 	CODE HERE;
|> END;
|> 
|> Now the above program (its in an implementation module)
|> compiles and i have no problem linking it to the client
|> program and the client program calls the above OK.
|> 
|> Now (here's the funny bit), if i remove both the above
|> WriteString procedures, (and re-compile, re-link), then
|> it crashes and doesn't seem to call the procedure (almost
|> like it's skipping it).
|> 
|> Does ANYONE have any idea what's wrong ?
|> 
|> Is there a MAJOR bug is Modula-2 ? I am using GPM September '95.

I have no idea what in particular might be wrong but whenever you add or
remove code you change the way it and it's data may lie in memory.  You
could very well have some uninitialized data that gets a valid value when
the Write's are there but gets a "bad" value when they are not.  I see
this quite a lot in both my own code and in code I help others to debug.

- Bill Seurer     ID Tools and Compiler Development      IBM Rochester, MN
  Business: BillSeurer@vnet.ibm.com               Home: BillSeurer@aol.com

Are you using GPM-LINUX? GPM-PC?

I came across a similar problem late last year under GPM-LINUX. I had a piece
of code that would crash in certain procedures, so I added some WriteString's
to provide debug output. With the debugging code, however, the program would
not crash.

As far as I can remember, this is due to a bug in the way that GPM handles
procedures that accept parameters that are arrays of some structured type
(RECORDS, etc). The stack frame wasn't being set up correctly upon procedure
entry. However, using a few empty WriteString's (ie WriteString("")) seemed to
fix the problem. 

I spoke to the person that writes the GPM backend, and the bug was confirmed
and fixed. The latest GPM-Linux release should contain the bug fix.

Of course, this may not be the same bug that you are experiencing....

Cheers,

Geoff Elgey,
Researcher, Programming Languages and Systems Group,
Faculty of Information Technology,
QUT. Brisbane, Queensland, Australia.

I think to be any help we would need to know the compiler, memory model 
etc.

There are such things as compiler bugs, I've seen a few, especially in 
JPI V1.17. If you have the choice, turn off ALL optimisation, and try 
gain. Maybe change memory model too.

D.